linux/fs/afs/fs_probe.c

470 lines
12 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-or-later
/* AFS fileserver probing
*
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
* Copyright (C) 2018, 2020 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*/
#include <linux/sched.h>
#include <linux/slab.h>
#include "afs_fs.h"
#include "internal.h"
#include "protocol_yfs.h"
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
static unsigned int afs_fs_probe_fast_poll_interval = 30 * HZ;
static unsigned int afs_fs_probe_slow_poll_interval = 5 * 60 * HZ;
/*
* Start the probe polling timer. We have to supply it with an inc on the
* outstanding server count.
*/
static void afs_schedule_fs_probe(struct afs_net *net,
struct afs_server *server, bool fast)
{
unsigned long atj;
if (!net->live)
return;
atj = server->probed_at;
atj += fast ? afs_fs_probe_fast_poll_interval : afs_fs_probe_slow_poll_interval;
afs_inc_servers_outstanding(net);
if (timer_reduce(&net->fs_probe_timer, atj))
afs_dec_servers_outstanding(net);
}
/*
* Handle the completion of a set of probes.
*/
static void afs_finished_fs_probe(struct afs_net *net, struct afs_server *server)
{
bool responded = server->probe.responded;
write_seqlock(&net->fs_lock);
if (responded) {
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
list_add_tail(&server->probe_link, &net->fs_probe_slow);
} else {
server->rtt = UINT_MAX;
clear_bit(AFS_SERVER_FL_RESPONDING, &server->flags);
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
list_add_tail(&server->probe_link, &net->fs_probe_fast);
}
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
write_sequnlock(&net->fs_lock);
afs_schedule_fs_probe(net, server, !responded);
}
/*
* Handle the completion of a probe.
*/
static void afs_done_one_fs_probe(struct afs_net *net, struct afs_server *server)
{
_enter("");
if (atomic_dec_and_test(&server->probe_outstanding))
afs_finished_fs_probe(net, server);
wake_up_all(&server->probe_wq);
}
/*
* Handle inability to send a probe due to ENOMEM when trying to allocate a
* call struct.
*/
static void afs_fs_probe_not_done(struct afs_net *net,
struct afs_server *server,
struct afs_addr_cursor *ac)
{
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
struct afs_addr_list *alist = ac->alist;
unsigned int index = ac->index;
_enter("");
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
trace_afs_io_error(0, -ENOMEM, afs_io_error_fs_probe_fail);
spin_lock(&server->probe_lock);
server->probe.local_failure = true;
if (server->probe.error == 0)
server->probe.error = -ENOMEM;
set_bit(index, &alist->failed);
spin_unlock(&server->probe_lock);
return afs_done_one_fs_probe(net, server);
}
/*
* Process the result of probing a fileserver. This is called after successful
* or failed delivery of an FS.GetCapabilities operation.
*/
void afs_fileserver_probe_result(struct afs_call *call)
{
struct afs_addr_list *alist = call->alist;
struct afs_server *server = call->server;
unsigned int index = call->addr_ix;
unsigned int rtt_us = 0;
int ret = call->error;
_enter("%pU,%u", &server->uuid, index);
spin_lock(&server->probe_lock);
switch (ret) {
case 0:
server->probe.error = 0;
goto responded;
case -ECONNABORTED:
if (!server->probe.responded) {
server->probe.abort_code = call->abort_code;
server->probe.error = ret;
}
goto responded;
case -ENOMEM:
case -ENONET:
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
clear_bit(index, &alist->responded);
server->probe.local_failure = true;
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
trace_afs_io_error(call->debug_id, ret, afs_io_error_fs_probe_fail);
goto out;
case -ECONNRESET: /* Responded, but call expired. */
case -ERFKILL:
case -EADDRNOTAVAIL:
case -ENETUNREACH:
case -EHOSTUNREACH:
case -EHOSTDOWN:
case -ECONNREFUSED:
case -ETIMEDOUT:
case -ETIME:
default:
clear_bit(index, &alist->responded);
set_bit(index, &alist->failed);
if (!server->probe.responded &&
(server->probe.error == 0 ||
server->probe.error == -ETIMEDOUT ||
server->probe.error == -ETIME))
server->probe.error = ret;
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
trace_afs_io_error(call->debug_id, ret, afs_io_error_fs_probe_fail);
goto out;
}
responded:
clear_bit(index, &alist->failed);
if (call->service_id == YFS_FS_SERVICE) {
server->probe.is_yfs = true;
set_bit(AFS_SERVER_FL_IS_YFS, &server->flags);
alist->addrs[index].srx_service = call->service_id;
} else {
server->probe.not_yfs = true;
if (!server->probe.is_yfs) {
clear_bit(AFS_SERVER_FL_IS_YFS, &server->flags);
alist->addrs[index].srx_service = call->service_id;
}
}
rxrpc: Fix the excessive initial retransmission timeout rxrpc currently uses a fixed 4s retransmission timeout until the RTT is sufficiently sampled. This can cause problems with some fileservers with calls to the cache manager in the afs filesystem being dropped from the fileserver because a packet goes missing and the retransmission timeout is greater than the call expiry timeout. Fix this by: (1) Copying the RTT/RTO calculation code from Linux's TCP implementation and altering it to fit rxrpc. (2) Altering the various users of the RTT to make use of the new SRTT value. (3) Replacing the use of rxrpc_resend_timeout to use the calculated RTO value instead (which is needed in jiffies), along with a backoff. Notes: (1) rxrpc provides RTT samples by matching the serial numbers on outgoing DATA packets that have the RXRPC_REQUEST_ACK set and PING ACK packets against the reference serial number in incoming REQUESTED ACK and PING-RESPONSE ACK packets. (2) Each packet that is transmitted on an rxrpc connection gets a new per-connection serial number, even for retransmissions, so an ACK can be cross-referenced to a specific trigger packet. This allows RTT information to be drawn from retransmitted DATA packets also. (3) rxrpc maintains the RTT/RTO state on the rxrpc_peer record rather than on an rxrpc_call because many RPC calls won't live long enough to generate more than one sample. (4) The calculated SRTT value is in units of 8ths of a microsecond rather than nanoseconds. The (S)RTT and RTO values are displayed in /proc/net/rxrpc/peers. Fixes: 17926a79320a ([AF_RXRPC]: Provide secure RxRPC sockets for use by userspace and kernel both"") Signed-off-by: David Howells <dhowells@redhat.com>
2020-05-11 21:54:34 +08:00
rtt_us = rxrpc_kernel_get_srtt(call->net->socket, call->rxcall);
if (rtt_us < server->probe.rtt) {
server->probe.rtt = rtt_us;
server->rtt = rtt_us;
alist->preferred = index;
}
smp_wmb(); /* Set rtt before responded. */
server->probe.responded = true;
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
set_bit(index, &alist->responded);
set_bit(AFS_SERVER_FL_RESPONDING, &server->flags);
out:
spin_unlock(&server->probe_lock);
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
_debug("probe %pU [%u] %pISpc rtt=%u ret=%d",
&server->uuid, index, &alist->addrs[index].transport,
rtt_us, ret);
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
return afs_done_one_fs_probe(call->net, server);
}
/*
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
* Probe one or all of a fileserver's addresses to find out the best route and
* to query its capabilities.
*/
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
void afs_fs_probe_fileserver(struct afs_net *net, struct afs_server *server,
struct key *key, bool all)
{
struct afs_addr_cursor ac = {
.index = 0,
};
_enter("%pU", &server->uuid);
read_lock(&server->fs_lock);
ac.alist = rcu_dereference_protected(server->addresses,
lockdep_is_held(&server->fs_lock));
afs_get_addrlist(ac.alist);
read_unlock(&server->fs_lock);
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
server->probed_at = jiffies;
atomic_set(&server->probe_outstanding, all ? ac.alist->nr_addrs : 1);
memset(&server->probe, 0, sizeof(server->probe));
server->probe.rtt = UINT_MAX;
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
ac.index = ac.alist->preferred;
if (ac.index < 0 || ac.index >= ac.alist->nr_addrs)
all = true;
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
if (all) {
for (ac.index = 0; ac.index < ac.alist->nr_addrs; ac.index++)
if (!afs_fs_get_capabilities(net, server, &ac, key))
afs_fs_probe_not_done(net, server, &ac);
} else {
if (!afs_fs_get_capabilities(net, server, &ac, key))
afs_fs_probe_not_done(net, server, &ac);
}
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
afs_put_addrlist(ac.alist);
}
/*
* Wait for the first as-yet untried fileserver to respond.
*/
int afs_wait_for_fs_probes(struct afs_server_list *slist, unsigned long untried)
{
struct wait_queue_entry *waits;
struct afs_server *server;
unsigned int rtt = UINT_MAX, rtt_s;
bool have_responders = false;
int pref = -1, i;
_enter("%u,%lx", slist->nr_servers, untried);
/* Only wait for servers that have a probe outstanding. */
for (i = 0; i < slist->nr_servers; i++) {
if (test_bit(i, &untried)) {
server = slist->servers[i].server;
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
if (!atomic_read(&server->probe_outstanding))
__clear_bit(i, &untried);
if (server->probe.responded)
have_responders = true;
}
}
if (have_responders || !untried)
return 0;
waits = kmalloc(array_size(slist->nr_servers, sizeof(*waits)), GFP_KERNEL);
if (!waits)
return -ENOMEM;
for (i = 0; i < slist->nr_servers; i++) {
if (test_bit(i, &untried)) {
server = slist->servers[i].server;
init_waitqueue_entry(&waits[i], current);
add_wait_queue(&server->probe_wq, &waits[i]);
}
}
for (;;) {
bool still_probing = false;
set_current_state(TASK_INTERRUPTIBLE);
for (i = 0; i < slist->nr_servers; i++) {
if (test_bit(i, &untried)) {
server = slist->servers[i].server;
if (server->probe.responded)
goto stop;
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
if (atomic_read(&server->probe_outstanding))
still_probing = true;
}
}
if (!still_probing || signal_pending(current))
goto stop;
schedule();
}
stop:
set_current_state(TASK_RUNNING);
for (i = 0; i < slist->nr_servers; i++) {
if (test_bit(i, &untried)) {
server = slist->servers[i].server;
rtt_s = READ_ONCE(server->rtt);
if (test_bit(AFS_SERVER_FL_RESPONDING, &server->flags) &&
rtt_s < rtt) {
pref = i;
rtt = rtt_s;
}
remove_wait_queue(&server->probe_wq, &waits[i]);
}
}
kfree(waits);
if (pref == -1 && signal_pending(current))
return -ERESTARTSYS;
if (pref >= 0)
slist->preferred = pref;
return 0;
}
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
/*
* Probe timer. We have an increment on fs_outstanding that we need to pass
* along to the work item.
*/
void afs_fs_probe_timer(struct timer_list *timer)
{
struct afs_net *net = container_of(timer, struct afs_net, fs_probe_timer);
if (!net->live || !queue_work(afs_wq, &net->fs_prober))
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
afs_dec_servers_outstanding(net);
}
/*
* Dispatch a probe to a server.
*/
static void afs_dispatch_fs_probe(struct afs_net *net, struct afs_server *server, bool all)
__releases(&net->fs_lock)
{
struct key *key = NULL;
/* We remove it from the queues here - it will be added back to
* one of the queues on the completion of the probe.
*/
list_del_init(&server->probe_link);
afs_get_server(server, afs_server_trace_get_probe);
write_sequnlock(&net->fs_lock);
afs_fs_probe_fileserver(net, server, key, all);
afs_put_server(net, server, afs_server_trace_put_probe);
}
/*
* Probe a server immediately without waiting for its due time to come
* round. This is used when all of the addresses have been tried.
*/
void afs_probe_fileserver(struct afs_net *net, struct afs_server *server)
{
write_seqlock(&net->fs_lock);
if (!list_empty(&server->probe_link))
return afs_dispatch_fs_probe(net, server, true);
write_sequnlock(&net->fs_lock);
}
afs: Actively poll fileservers to maintain NAT or firewall openings When an AFS client accesses a file, it receives a limited-duration callback promise that the server will notify it if another client changes a file. This callback duration can be a few hours in length. If a client mounts a volume and then an application prevents it from being unmounted, say by chdir'ing into it, but then does nothing for some time, the rxrpc_peer record will expire and rxrpc-level keepalive will cease. If there is NAT or a firewall between the client and the server, the route back for the server may close after a comparatively short duration, meaning that attempts by the server to notify the client may then bounce. The client, however, may (so far as it knows) still have a valid unexpired promise and will then rely on its cached data and will not see changes made on the server by a third party until it incidentally rechecks the status or the promise needs renewal. To deal with this, the client needs to regularly probe the server. This has two effects: firstly, it keeps a route open back for the server, and secondly, it causes the server to disgorge any notifications that got queued up because they couldn't be sent. Fix this by adding a mechanism to emit regular probes. Two levels of probing are made available: Under normal circumstances the 'slow' queue will be used for a fileserver - this just probes the preferred address once every 5 mins or so; however, if server fails to respond to any probes, the server will shift to the 'fast' queue from which all its interfaces will be probed every 30s. When it finally responds, the record will switch back to the slow queue. Further notes: (1) Probing is now no longer driven from the fileserver rotation algorithm. (2) Probes are dispatched to all interfaces on a fileserver when that an afs_server object is set up to record it. (3) The afs_server object is removed from the probe queues when we start to probe it. afs_is_probing_server() returns true if it's not listed - ie. it's undergoing probing. (4) The afs_server object is added back on to the probe queue when the final outstanding probe completes, but the probed_at time is set when we're about to launch a probe so that it's not dependent on the probe duration. (5) The timer and the work item added for this must be handed a count on net->servers_outstanding, which they hand on or release. This makes sure that network namespace cleanup waits for them. Fixes: d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation") Reported-by: Dave Botsch <botsch@cnf.cornell.edu> Signed-off-by: David Howells <dhowells@redhat.com>
2020-04-24 22:10:00 +08:00
/*
* Probe dispatcher to regularly dispatch probes to keep NAT alive.
*/
void afs_fs_probe_dispatcher(struct work_struct *work)
{
struct afs_net *net = container_of(work, struct afs_net, fs_prober);
struct afs_server *fast, *slow, *server;
unsigned long nowj, timer_at, poll_at;
bool first_pass = true, set_timer = false;
if (!net->live)
return;
_enter("");
if (list_empty(&net->fs_probe_fast) && list_empty(&net->fs_probe_slow)) {
_leave(" [none]");
return;
}
again:
write_seqlock(&net->fs_lock);
fast = slow = server = NULL;
nowj = jiffies;
timer_at = nowj + MAX_JIFFY_OFFSET;
if (!list_empty(&net->fs_probe_fast)) {
fast = list_first_entry(&net->fs_probe_fast, struct afs_server, probe_link);
poll_at = fast->probed_at + afs_fs_probe_fast_poll_interval;
if (time_before(nowj, poll_at)) {
timer_at = poll_at;
set_timer = true;
fast = NULL;
}
}
if (!list_empty(&net->fs_probe_slow)) {
slow = list_first_entry(&net->fs_probe_slow, struct afs_server, probe_link);
poll_at = slow->probed_at + afs_fs_probe_slow_poll_interval;
if (time_before(nowj, poll_at)) {
if (time_before(poll_at, timer_at))
timer_at = poll_at;
set_timer = true;
slow = NULL;
}
}
server = fast ?: slow;
if (server)
_debug("probe %pU", &server->uuid);
if (server && (first_pass || !need_resched())) {
afs_dispatch_fs_probe(net, server, server == fast);
first_pass = false;
goto again;
}
write_sequnlock(&net->fs_lock);
if (server) {
if (!queue_work(afs_wq, &net->fs_prober))
afs_dec_servers_outstanding(net);
_leave(" [requeue]");
} else if (set_timer) {
if (timer_reduce(&net->fs_probe_timer, timer_at))
afs_dec_servers_outstanding(net);
_leave(" [timer]");
} else {
afs_dec_servers_outstanding(net);
_leave(" [quiesce]");
}
}
/*
* Wait for a probe on a particular fileserver to complete for 2s.
*/
int afs_wait_for_one_fs_probe(struct afs_server *server, bool is_intr)
{
struct wait_queue_entry wait;
unsigned long timo = 2 * HZ;
if (atomic_read(&server->probe_outstanding) == 0)
goto dont_wait;
init_wait_entry(&wait, 0);
for (;;) {
prepare_to_wait_event(&server->probe_wq, &wait,
is_intr ? TASK_INTERRUPTIBLE : TASK_UNINTERRUPTIBLE);
if (timo == 0 ||
server->probe.responded ||
atomic_read(&server->probe_outstanding) == 0 ||
(is_intr && signal_pending(current)))
break;
timo = schedule_timeout(timo);
}
finish_wait(&server->probe_wq, &wait);
dont_wait:
if (server->probe.responded)
return 0;
if (is_intr && signal_pending(current))
return -ERESTARTSYS;
if (timo == 0)
return -ETIME;
return -EDESTADDRREQ;
}
/*
* Clean up the probing when the namespace is killed off.
*/
void afs_fs_probe_cleanup(struct afs_net *net)
{
if (del_timer_sync(&net->fs_probe_timer))
afs_dec_servers_outstanding(net);
}