diff options
author | David Howells <dhowells@redhat.com> | 2018-10-20 00:57:59 +0100 |
---|---|---|
committer | David Howells <dhowells@redhat.com> | 2018-10-24 00:41:09 +0100 |
commit | 3bf0fb6f33dd545693da5e65f5b1b9b9f0bfc35e (patch) | |
tree | df215e6a6ad11b6ac8158461144667e168591d28 /fs/afs/cmservice.c | |
parent | 18ac61853cc4e44eb30e125fc8344a3b25c7b6fe (diff) |
afs: Probe multiple fileservers simultaneously
Send probes to all the unprobed fileservers in a fileserver list on all
addresses simultaneously in an attempt to find out the fastest route whilst
not getting stuck for 20s on any server or address that we don't get a
reply from.
This alleviates the problem whereby attempting to access a new server can
take a long time because the rotation algorithm ends up rotating through
all servers and addresses until it finds one that responds.
Signed-off-by: David Howells <dhowells@redhat.com>
Diffstat (limited to 'fs/afs/cmservice.c')
-rw-r--r-- | fs/afs/cmservice.c | 129 |
1 files changed, 94 insertions, 35 deletions
diff --git a/fs/afs/cmservice.c b/fs/afs/cmservice.c index 8cf8d10daa6c..8ee5972893ed 100644 --- a/fs/afs/cmservice.c +++ b/fs/afs/cmservice.c @@ -122,6 +122,8 @@ bool afs_cm_incoming_call(struct afs_call *call) { _enter("{%u, CB.OP %u}", call->service_id, call->operation_ID); + call->epoch = rxrpc_kernel_get_epoch(call->net->socket, call->rxcall); + switch (call->operation_ID) { case CBCallBack: call->type = &afs_SRXCBCallBack; @@ -152,6 +154,91 @@ bool afs_cm_incoming_call(struct afs_call *call) } /* + * Record a probe to the cache manager from a server. + */ +static int afs_record_cm_probe(struct afs_call *call, struct afs_server *server) +{ + _enter(""); + + if (test_bit(AFS_SERVER_FL_HAVE_EPOCH, &server->flags) && + !test_bit(AFS_SERVER_FL_PROBING, &server->flags)) { + if (server->cm_epoch == call->epoch) + return 0; + + if (!server->probe.said_rebooted) { + pr_notice("kAFS: FS rebooted %pU\n", &server->uuid); + server->probe.said_rebooted = true; + } + } + + spin_lock(&server->probe_lock); + + if (!test_bit(AFS_SERVER_FL_HAVE_EPOCH, &server->flags)) { + server->cm_epoch = call->epoch; + server->probe.cm_epoch = call->epoch; + goto out; + } + + if (server->probe.cm_probed && + call->epoch != server->probe.cm_epoch && + !server->probe.said_inconsistent) { + pr_notice("kAFS: FS endpoints inconsistent %pU\n", + &server->uuid); + server->probe.said_inconsistent = true; + } + + if (!server->probe.cm_probed || call->epoch == server->cm_epoch) + server->probe.cm_epoch = server->cm_epoch; + +out: + server->probe.cm_probed = true; + spin_unlock(&server->probe_lock); + return 0; +} + +/* + * Find the server record by peer address and record a probe to the cache + * manager from a server. + */ +static int afs_find_cm_server_by_peer(struct afs_call *call) +{ + struct sockaddr_rxrpc srx; + struct afs_server *server; + + rxrpc_kernel_get_peer(call->net->socket, call->rxcall, &srx); + + server = afs_find_server(call->net, &srx); + if (!server) { + trace_afs_cm_no_server(call, &srx); + return 0; + } + + call->cm_server = server; + return afs_record_cm_probe(call, server); +} + +/* + * Find the server record by server UUID and record a probe to the cache + * manager from a server. + */ +static int afs_find_cm_server_by_uuid(struct afs_call *call, + struct afs_uuid *uuid) +{ + struct afs_server *server; + + rcu_read_lock(); + server = afs_find_server_by_uuid(call->net, call->request); + rcu_read_unlock(); + if (!server) { + trace_afs_cm_no_server_u(call, call->request); + return 0; + } + + call->cm_server = server; + return afs_record_cm_probe(call, server); +} + +/* * Clean up a cache manager call. */ static void afs_cm_destructor(struct afs_call *call) @@ -187,7 +274,6 @@ static void SRXAFSCB_CallBack(struct work_struct *work) static int afs_deliver_cb_callback(struct afs_call *call) { struct afs_callback_break *cb; - struct sockaddr_rxrpc srx; __be32 *bp; int ret, loop; @@ -276,12 +362,7 @@ static int afs_deliver_cb_callback(struct afs_call *call) /* we'll need the file server record as that tells us which set of * vnodes to operate upon */ - rxrpc_kernel_get_peer(call->net->socket, call->rxcall, &srx); - call->cm_server = afs_find_server(call->net, &srx); - if (!call->cm_server) - trace_afs_cm_no_server(call, &srx); - - return afs_queue_call_work(call); + return afs_find_cm_server_by_peer(call); } /* @@ -305,13 +386,10 @@ static void SRXAFSCB_InitCallBackState(struct work_struct *work) */ static int afs_deliver_cb_init_call_back_state(struct afs_call *call) { - struct sockaddr_rxrpc srx; int ret; _enter(""); - rxrpc_kernel_get_peer(call->net->socket, call->rxcall, &srx); - afs_extract_discard(call, 0); ret = afs_extract_data(call, false); if (ret < 0) @@ -319,11 +397,7 @@ static int afs_deliver_cb_init_call_back_state(struct afs_call *call) /* we'll need the file server record as that tells us which set of * vnodes to operate upon */ - call->cm_server = afs_find_server(call->net, &srx); - if (!call->cm_server) - trace_afs_cm_no_server(call, &srx); - - return afs_queue_call_work(call); + return afs_find_cm_server_by_peer(call); } /* @@ -384,13 +458,7 @@ static int afs_deliver_cb_init_call_back_state3(struct afs_call *call) /* we'll need the file server record as that tells us which set of * vnodes to operate upon */ - rcu_read_lock(); - call->cm_server = afs_find_server_by_uuid(call->net, call->request); - rcu_read_unlock(); - if (!call->cm_server) - trace_afs_cm_no_server_u(call, call->request); - - return afs_queue_call_work(call); + return afs_find_cm_server_by_uuid(call, call->request); } /* @@ -422,8 +490,7 @@ static int afs_deliver_cb_probe(struct afs_call *call) if (!afs_check_call_state(call, AFS_CALL_SV_REPLYING)) return afs_io_error(call, afs_io_error_cm_reply); - - return afs_queue_call_work(call); + return afs_find_cm_server_by_peer(call); } /* @@ -503,8 +570,7 @@ static int afs_deliver_cb_probe_uuid(struct afs_call *call) if (!afs_check_call_state(call, AFS_CALL_SV_REPLYING)) return afs_io_error(call, afs_io_error_cm_reply); - - return afs_queue_call_work(call); + return afs_find_cm_server_by_uuid(call, call->request); } /* @@ -586,8 +652,7 @@ static int afs_deliver_cb_tell_me_about_yourself(struct afs_call *call) if (!afs_check_call_state(call, AFS_CALL_SV_REPLYING)) return afs_io_error(call, afs_io_error_cm_reply); - - return afs_queue_call_work(call); + return afs_find_cm_server_by_peer(call); } /* @@ -596,7 +661,6 @@ static int afs_deliver_cb_tell_me_about_yourself(struct afs_call *call) static int afs_deliver_yfs_cb_callback(struct afs_call *call) { struct afs_callback_break *cb; - struct sockaddr_rxrpc srx; struct yfs_xdr_YFSFid *bp; size_t size; int ret, loop; @@ -664,10 +728,5 @@ static int afs_deliver_yfs_cb_callback(struct afs_call *call) /* We'll need the file server record as that tells us which set of * vnodes to operate upon. */ - rxrpc_kernel_get_peer(call->net->socket, call->rxcall, &srx); - call->cm_server = afs_find_server(call->net, &srx); - if (!call->cm_server) - trace_afs_cm_no_server(call, &srx); - - return afs_queue_call_work(call); + return afs_find_cm_server_by_peer(call); } |