X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Fptlrpc%2Fpinger.c;h=7a15386f72e5aafb7e246e20b841c675153597f6;hb=c86481c7290dffd7ed579d7c8d8ac8676d8c4d79;hp=819cad6d6a314fe7d1ec88a0a9fe64d01fb00db6;hpb=b61d3781d58689a994f4687d9396db17386926bd;p=fs%2Flustre-release.git diff --git a/lustre/ptlrpc/pinger.c b/lustre/ptlrpc/pinger.c index 819cad6..7a15386 100644 --- a/lustre/ptlrpc/pinger.c +++ b/lustre/ptlrpc/pinger.c @@ -1,29 +1,41 @@ /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- * vim:expandtab:shiftwidth=8:tabstop=8: * - * Portal-RPC reconnection and replay operations, for use in recovery. + * GPL HEADER START + * + * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * - * Copyright (c) 2003 Cluster File Systems, Inc. - * Authors: Phil Schwan - * Mike Shaver + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 only, + * as published by the Free Software Foundation. * - * This file is part of the Lustre file system, http://www.lustre.org - * Lustre is a trademark of Cluster File Systems, Inc. + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License version 2 for more details (a copy is included + * in the LICENSE file that accompanied this code). * - * You may have signed or agreed to another license before downloading - * this software. If so, you are bound by the terms and conditions - * of that agreement, and the following does not apply to you. See the - * LICENSE file included with this distribution for more information. + * You should have received a copy of the GNU General Public License + * version 2 along with this program; If not, see + * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf * - * If you did not agree to a different license, then this copy of Lustre - * is open source software; you can redistribute it and/or modify it - * under the terms of version 2 of the GNU General Public License as - * published by the Free Software Foundation. + * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, + * CA 95054 USA or visit www.sun.com if you need additional information or + * have any questions. * - * In either case, Lustre is distributed in the hope that it will be - * useful, but WITHOUT ANY WARRANTY; without even the implied warranty - * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * license text for more details. + * GPL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved + * Use is subject to license terms. + */ +/* + * This file is part of Lustre, http://www.lustre.org/ + * Lustre is a trademark of Sun Microsystems, Inc. + * + * lustre/ptlrpc/pinger.c + * + * Portal-RPC reconnection and replay operations, for use in recovery. */ #ifndef __KERNEL__ @@ -37,15 +49,48 @@ #include "ptlrpc_internal.h" struct semaphore pinger_sem; -static struct list_head pinger_imports = CFS_LIST_HEAD_INIT(pinger_imports); +static CFS_LIST_HEAD(pinger_imports); + +struct ptlrpc_request * +ptlrpc_prep_ping(struct obd_import *imp) +{ + struct ptlrpc_request *req; + + req = ptlrpc_request_alloc_pack(imp, &RQF_OBD_PING, + LUSTRE_OBD_VERSION, OBD_PING); + if (req) { + ptlrpc_request_set_replen(req); + req->rq_no_resend = req->rq_no_delay = 1; + } + return req; +} + +int ptlrpc_obd_ping(struct obd_device *obd) +{ + int rc; + struct ptlrpc_request *req; + ENTRY; + + req = ptlrpc_prep_ping(obd->u.cli.cl_import); + if (req == NULL) + RETURN(-ENOMEM); + + req->rq_send_state = LUSTRE_IMP_FULL; + + rc = ptlrpc_queue_wait(req); + + ptlrpc_req_finished(req); + + RETURN(rc); +} +EXPORT_SYMBOL(ptlrpc_obd_ping); int ptlrpc_ping(struct obd_import *imp) { struct ptlrpc_request *req; ENTRY; - req = ptlrpc_request_alloc_pack(imp, &RQF_OBD_PING, LUSTRE_OBD_VERSION, - OBD_PING); + req = ptlrpc_prep_ping(imp); if (req == NULL) { CERROR("OOM trying to ping %s->%s\n", imp->imp_obd->obd_uuid.uuid, @@ -55,10 +100,7 @@ int ptlrpc_ping(struct obd_import *imp) DEBUG_REQ(D_INFO, req, "pinging %s->%s", imp->imp_obd->obd_uuid.uuid, obd2cli_tgt(imp->imp_obd)); - req->rq_no_resend = req->rq_no_delay = 1; - ptlrpc_request_set_replen(req); - req->rq_timeout = PING_INTERVAL; - ptlrpcd_add_req(req); + ptlrpcd_add_req(req, PSCOPE_OTHER); RETURN(0); } @@ -66,9 +108,14 @@ int ptlrpc_ping(struct obd_import *imp) void ptlrpc_update_next_ping(struct obd_import *imp) { #ifdef ENABLE_PINGER - imp->imp_next_ping = cfs_time_shift( - (imp->imp_state == LUSTRE_IMP_DISCON ? - RECONNECT_INTERVAL : PING_INTERVAL)); + int time = PING_INTERVAL; + if (imp->imp_state == LUSTRE_IMP_DISCON) { + int dtime = max_t(int, CONNECTION_SWITCH_MIN, + AT_OFF ? 0 : + at_get(&imp->imp_at.iat_net_latency)); + time = min(time, dtime); + } + imp->imp_next_ping = cfs_time_shift(time); #endif /* ENABLE_PINGER */ } @@ -175,7 +222,7 @@ static void ptlrpc_pinger_process_import(struct obd_import *imp, /* wait at least a timeout before trying recovery again */ imp->imp_next_ping = ptlrpc_next_reconnect(imp); ptlrpc_initiate_recovery(imp); - } else if (level != LUSTRE_IMP_FULL || + } else if (level != LUSTRE_IMP_FULL || imp->imp_obd->obd_no_recov || imp->imp_deactive) { CDEBUG(D_HA, "not pinging %s (in recovery " @@ -215,8 +262,8 @@ static int ptlrpc_pinger_main(void *arg) ptlrpc_pinger_process_import(imp, this_ping); /* obd_timeout might have changed */ if (imp->imp_pingable && imp->imp_next_ping && - cfs_time_after(imp->imp_next_ping, - cfs_time_add(this_ping, + cfs_time_after(imp->imp_next_ping, + cfs_time_add(this_ping, cfs_time_seconds(PING_INTERVAL)))) ptlrpc_update_next_ping(imp); } @@ -225,21 +272,24 @@ static int ptlrpc_pinger_main(void *arg) obd_update_maxusage(); /* Wait until the next ping time, or until we're stopped. */ - time_to_next_ping = cfs_time_sub(cfs_time_add(this_ping, - cfs_time_seconds(PING_INTERVAL)), - cfs_time_current()); + time_to_next_ping = cfs_time_sub(cfs_time_add(this_ping, + cfs_time_seconds(PING_INTERVAL)), + cfs_time_current()); /* The ping sent by ptlrpc_send_rpc may get sent out say .01 second after this. - ptlrpc_pinger_eending_on_import will then set the + ptlrpc_pinger_sending_on_import will then set the next ping time to next_ping + .01 sec, which means we will SKIP the next ping at next_ping, and the ping will get sent 2 timeouts from now! Beware. */ - CDEBUG(D_INFO, "next ping in "CFS_DURATION_T" ("CFS_TIME_T")\n", - time_to_next_ping, - cfs_time_add(this_ping, cfs_time_seconds(PING_INTERVAL))); + CDEBUG(D_INFO, "next ping in "CFS_DURATION_T" ("CFS_TIME_T")\n", + time_to_next_ping, + cfs_time_add(this_ping, + cfs_time_seconds(PING_INTERVAL))); if (time_to_next_ping > 0) { - lwi = LWI_TIMEOUT(max_t(cfs_duration_t, time_to_next_ping, cfs_time_seconds(1)), + lwi = LWI_TIMEOUT(max_t(cfs_duration_t, + time_to_next_ping, + cfs_time_seconds(1)), NULL, NULL); l_wait_event(thread->t_ctl_waitq, thread->t_flags & (SVC_STOPPING|SVC_EVENT), @@ -282,7 +332,7 @@ int ptlrpc_start_pinger(void) RETURN(-ENOMEM); cfs_waitq_init(&pinger_thread->t_ctl_waitq); cfs_waitq_init(&suspend_timeouts_waitq); - + d.name = "ll_ping"; d.thread = pinger_thread; @@ -433,7 +483,7 @@ static int ping_evictor_main(void *arg) obd = pet_exp->exp_obd; spin_unlock(&pet_lock); - expire_time = cfs_time_current_sec() - (3 * obd_timeout / 2); + expire_time = cfs_time_current_sec() - PING_EVICT_TIMEOUT; CDEBUG(D_HA, "evicting all exports of obd %s older than %ld\n", obd->obd_name, expire_time); @@ -531,6 +581,7 @@ static int pinger_check_rpcs(void *arg) struct ptlrpc_request *req; struct ptlrpc_request_set *set; struct list_head *iter; + struct obd_import *imp; struct pinger_data *pd = &pinger_args; int rc; @@ -568,7 +619,7 @@ static int pinger_check_rpcs(void *arg) list_entry(iter, struct obd_import, imp_pinger_chain); int generation, level; - if (cfs_time_aftereq(pd->pd_this_ping, + if (cfs_time_aftereq(pd->pd_this_ping, imp->imp_next_ping - 5 * CFS_TICK)) { /* Add a ping. */ spin_lock(&imp->imp_lock); @@ -596,7 +647,7 @@ static int pinger_check_rpcs(void *arg) req->rq_no_resend = 1; ptlrpc_request_set_replen(req); req->rq_send_state = LUSTRE_IMP_FULL; - req->rq_phase = RQ_PHASE_RPC; + ptlrpc_rqphase_move(req, RQ_PHASE_RPC); req->rq_import_generation = generation; ptlrpc_set_add_req(set, req); } else { @@ -623,7 +674,7 @@ static int pinger_check_rpcs(void *arg) } do_check_set: - rc = ptlrpc_check_set(set); + rc = ptlrpc_check_set(NULL, set); /* not finished, and we are not expired, simply return */ if (!rc && cfs_time_before(curtime, cfs_time_add(pd->pd_this_ping, @@ -642,17 +693,23 @@ do_check_set: if (req->rq_phase == RQ_PHASE_COMPLETE) continue; - req->rq_phase = RQ_PHASE_COMPLETE; - atomic_dec(&req->rq_import->imp_inflight); - set->set_remaining--; - /* If it was disconnected, don't sweat it. */ - if (list_empty(&req->rq_import->imp_pinger_chain)) { - ptlrpc_unregister_reply(req); - continue; - } + CDEBUG(D_RPCTRACE, "Pinger initiate expire request(%p)\n", + req); + + /* This will also unregister reply. */ + ptlrpc_expire_one_request(req, 0); - CDEBUG(D_RPCTRACE, "pinger initiate expire_one_request\n"); - ptlrpc_expire_one_request(req); + /* We're done with this req, let's finally move it to complete + * phase and take care of inflights. */ + ptlrpc_rqphase_move(req, RQ_PHASE_COMPLETE); + imp = req->rq_import; + spin_lock(&imp->imp_lock); + if (!list_empty(&req->rq_list)) { + list_del_init(&req->rq_list); + atomic_dec(&imp->imp_inflight); + } + spin_unlock(&imp->imp_lock); + set->set_remaining--; } mutex_up(&pinger_sem);