/* -*- 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. * * Copyright (c) 2002, 2003 Cluster File Systems, Inc. * Author: Mike Shaver * * This file is part of Lustre, http://www.lustre.org. * * Lustre is free 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. * * 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 * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with Lustre; if not, write to the Free Software * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ #define DEBUG_SUBSYSTEM S_RPC #ifdef __KERNEL__ # include # include # include #else # include #endif #include #include #include #include #include #include #include #include #include /* for IOC_LOV_SET_OSC_ACTIVE */ #include "ptlrpc_internal.h" static int ptlrpc_recover_import_no_retry(struct obd_import *, char *); void ptlrpc_run_recovery_over_upcall(struct obd_device *obd) { char *argv[4]; char *envp[3]; int rc; ENTRY; argv[0] = obd_lustre_upcall; argv[1] = "RECOVERY_OVER"; argv[2] = obd->obd_uuid.uuid; argv[3] = NULL; envp[0] = "HOME=/"; envp[1] = "PATH=/sbin:/bin:/usr/sbin:/usr/bin"; envp[2] = NULL; rc = USERMODEHELPER(argv[0], argv, envp); if (rc < 0) { CERROR("Error invoking recovery upcall %s %s %s: %d; check " "/proc/sys/lustre/upcall\n", argv[0], argv[1], argv[2], rc); } else { CERROR("Invoked upcall %s %s %s", argv[0], argv[1], argv[2]); } } void ptlrpc_run_failed_import_upcall(struct obd_import* imp) { #ifdef __KERNEL__ char *argv[7]; char *envp[3]; int rc; ENTRY; argv[0] = obd_lustre_upcall; argv[1] = "FAILED_IMPORT"; argv[2] = imp->imp_target_uuid.uuid; argv[3] = imp->imp_obd->obd_name; argv[4] = imp->imp_connection->c_remote_uuid.uuid; argv[5] = imp->imp_obd->obd_uuid.uuid; argv[6] = NULL; envp[0] = "HOME=/"; envp[1] = "PATH=/sbin:/bin:/usr/sbin:/usr/bin"; envp[2] = NULL; rc = USERMODEHELPER(argv[0], argv, envp); if (rc < 0) { CERROR("Error invoking recovery upcall %s %s %s %s %s: %d; " "check /proc/sys/lustre/lustre_upcall\n", argv[0], argv[1], argv[2], argv[3], argv[4],rc); } else { CERROR("Invoked upcall %s %s %s %s %s\n", argv[0], argv[1], argv[2], argv[3], argv[4]); } #else ptlrpc_recover_import(imp, NULL); #endif } int ptlrpc_replay(struct obd_import *imp) { int rc = 0; struct list_head *tmp, *pos; struct ptlrpc_request *req; unsigned long flags; ENTRY; /* It might have committed some after we last spoke, so make sure we * get rid of them now. */ spin_lock_irqsave(&imp->imp_lock, flags); ptlrpc_free_committed(imp); spin_unlock_irqrestore(&imp->imp_lock, flags); CDEBUG(D_HA, "import %p from %s has committed "LPD64"\n", imp, imp->imp_target_uuid.uuid, imp->imp_peer_committed_transno); list_for_each(tmp, &imp->imp_replay_list) { req = list_entry(tmp, struct ptlrpc_request, rq_list); DEBUG_REQ(D_HA, req, "RETAINED: "); } /* Do I need to hold a lock across this iteration? We shouldn't be * racing with any additions to the list, because we're in recovery * and are therefore not processing additional requests to add. Calls * to ptlrpc_free_committed might commit requests, but nothing "newer" * than the one we're replaying (it can't be committed until it's * replayed, and we're doing that here). l_f_e_safe protects against * problems with the current request being committed, in the unlikely * event of that race. So, in conclusion, I think that it's safe to * perform this list-walk without the imp_lock held. * * But, the {mdc,osc}_replay_open callbacks both iterate * request lists, and have comments saying they assume the * imp_lock is being held by ptlrpc_replay, but it's not. it's * just a little race... */ list_for_each_safe(tmp, pos, &imp->imp_replay_list) { req = list_entry(tmp, struct ptlrpc_request, rq_list); DEBUG_REQ(D_HA, req, "REPLAY:"); rc = ptlrpc_replay_req(req); if (rc) { CERROR("recovery replay error %d for req "LPD64"\n", rc, req->rq_xid); RETURN(rc); } } RETURN(0); } int ptlrpc_resend(struct obd_import *imp) { struct list_head *tmp, *pos; struct ptlrpc_request *req; unsigned long flags; ENTRY; /* As long as we're in recovery, nothing should be added to the sending * list, so we don't need to hold the lock during this iteration and * resend process. */ /* Well... what if lctl recover is called twice at the same time? */ spin_lock_irqsave(&imp->imp_lock, flags); LASSERT(imp->imp_state == LUSTRE_IMP_RECOVER); spin_unlock_irqrestore(&imp->imp_lock, flags); list_for_each_safe(tmp, pos, &imp->imp_sending_list) { req = list_entry(tmp, struct ptlrpc_request, rq_list); ptlrpc_resend_req(req); } RETURN(0); } void ptlrpc_wake_delayed(struct obd_import *imp) { unsigned long flags; struct list_head *tmp, *pos; struct ptlrpc_request *req; spin_lock_irqsave(&imp->imp_lock, flags); list_for_each_safe(tmp, pos, &imp->imp_delayed_list) { req = list_entry(tmp, struct ptlrpc_request, rq_list); ptlrpc_put_connection(req->rq_connection); req->rq_connection = ptlrpc_connection_addref(req->rq_import->imp_connection); if (req->rq_set) { DEBUG_REQ(D_HA, req, "waking (set %p):", req->rq_set); wake_up(&req->rq_set->set_waitq); } else { DEBUG_REQ(D_HA, req, "waking:"); wake_up(&req->rq_reply_waitq); } } spin_unlock_irqrestore(&imp->imp_lock, flags); } inline void ptlrpc_invalidate_import_state(struct obd_import *imp) { struct obd_device *obd = imp->imp_obd; struct ldlm_namespace *ns = obd->obd_namespace; ptlrpc_abort_inflight(imp); #if 0 obd_invalidate_import(obd, imp); #endif ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY | LDLM_FL_CANCEL); } void ptlrpc_handle_failed_import(struct obd_import *imp) { ENTRY; if (!imp->imp_replayable) { CDEBUG(D_HA, "import %s@%s for %s not replayable, deactivating\n", imp->imp_target_uuid.uuid, imp->imp_connection->c_remote_uuid.uuid, imp->imp_obd->obd_name); ptlrpc_set_import_active(imp, 0); } ptlrpc_run_failed_import_upcall(imp); EXIT; } void ptlrpc_request_handle_notconn(struct ptlrpc_request *failed_req) { int rc; struct obd_import *imp= failed_req->rq_import; unsigned long flags; ENTRY; CDEBUG(D_HA, "import %s of %s@%s abruptly disconnected: reconnecting\n", imp->imp_obd->obd_name, imp->imp_target_uuid.uuid, imp->imp_connection->c_remote_uuid.uuid); rc = ptlrpc_recover_import_no_retry(imp, NULL); if (failed_req->rq_import_generation != imp->imp_generation) { spin_lock_irqsave (&failed_req->rq_lock, flags); failed_req->rq_err = 1; spin_unlock_irqrestore (&failed_req->rq_lock, flags); } else { ptlrpc_resend_req(failed_req); if (rc && rc != -EALREADY) ptlrpc_handle_failed_import(imp); } EXIT; } int ptlrpc_set_import_active(struct obd_import *imp, int active) { struct obd_device *obd = imp->imp_obd; unsigned long flags; LASSERT(obd); /* When deactivating, mark import invalid, and abort in-flight * requests. */ if (!active) { spin_lock_irqsave(&imp->imp_lock, flags); /* This is a bit of a hack, but invalidating replayable * imports makes a temporary reconnect failure into a much more * ugly -- and hard to remedy -- situation. */ if (!imp->imp_replayable) { CDEBUG(D_HA, "setting import %s INVALID\n", imp->imp_target_uuid.uuid); imp->imp_invalid = 1; } imp->imp_generation++; spin_unlock_irqrestore(&imp->imp_lock, flags); ptlrpc_invalidate_import_state(imp); } /* When activating, mark import valid */ if (active) { CDEBUG(D_HA, "setting import %s VALID\n", imp->imp_target_uuid.uuid); spin_lock_irqsave(&imp->imp_lock, flags); imp->imp_invalid = 0; spin_unlock_irqrestore(&imp->imp_lock, flags); } if (obd->obd_observer) RETURN(obd_notify(obd->obd_observer, obd, active)); RETURN(0); } void ptlrpc_fail_import(struct obd_import *imp, int generation) { unsigned long flags; int in_recovery = 0; ENTRY; LASSERT (!imp->imp_dlm_fake); spin_lock_irqsave(&imp->imp_lock, flags); if (imp->imp_state != LUSTRE_IMP_FULL) { in_recovery = 1; } else { CDEBUG(D_HA, "%s: new state: DISCON\n", imp->imp_client->cli_name); imp->imp_state = LUSTRE_IMP_DISCON; } spin_unlock_irqrestore(&imp->imp_lock, flags); if (in_recovery) { EXIT; return; } ptlrpc_handle_failed_import(imp); EXIT; } static int signal_completed_replay(struct obd_import *imp) { struct ptlrpc_request *req; int rc; ENTRY; req = ptlrpc_prep_req(imp, OBD_PING, 0, NULL, NULL); if (!req) RETURN(-ENOMEM); req->rq_replen = lustre_msg_size(0, NULL); req->rq_send_state = LUSTRE_IMP_REPLAY; req->rq_reqmsg->flags |= MSG_LAST_REPLAY; req->rq_timeout *= 3; rc = ptlrpc_queue_wait(req); ptlrpc_req_finished(req); RETURN(rc); } int ptlrpc_recover_import(struct obd_import *imp, char *new_uuid) { int rc; ENTRY; rc = ptlrpc_recover_import_no_retry(imp, new_uuid); if (rc && rc != -EALREADY) { unsigned long flags; CDEBUG(D_HA, "recovery of %s on %s failed (%d); restarting\n", imp->imp_target_uuid.uuid, new_uuid ? new_uuid : (char *)imp->imp_connection->c_remote_uuid.uuid, rc); spin_lock_irqsave(&imp->imp_lock, flags); imp->imp_state = LUSTRE_IMP_FULL; spin_unlock_irqrestore(&imp->imp_lock, flags); ptlrpc_fail_import(imp, imp->imp_generation); } RETURN(rc); } static int ptlrpc_recover_import_no_retry(struct obd_import *imp, char *new_uuid) { int rc; unsigned long flags; int in_recovery = 0; int was_invalid = 0; ENTRY; spin_lock_irqsave(&imp->imp_lock, flags); if (imp->imp_state == LUSTRE_IMP_FULL) { CDEBUG(D_HA, "%s: new state: DISCON\n", imp->imp_client->cli_name); imp->imp_state = LUSTRE_IMP_DISCON; } if (imp->imp_state != LUSTRE_IMP_DISCON) { in_recovery = 1; } else if (imp->imp_invalid) { imp->imp_invalid = 0; was_invalid = 1; } spin_unlock_irqrestore(&imp->imp_lock, flags); if (in_recovery == 1) RETURN(-EALREADY); down(&imp->imp_recovery_sem); /* If recovery happened while we waited, we're done. */ if (imp->imp_state == LUSTRE_IMP_FULL) GOTO(out, rc = 0); LASSERT (imp->imp_state == LUSTRE_IMP_DISCON); if (new_uuid) { struct ptlrpc_connection *conn; struct obd_uuid uuid; struct ptlrpc_peer peer; struct obd_export *dlmexp; obd_str2uuid(&uuid, new_uuid); if (ptlrpc_uuid_to_peer(&uuid, &peer)) { CERROR("no connection found for UUID %s\n", new_uuid); GOTO(out, rc = -EINVAL); } conn = ptlrpc_get_connection(&peer, &uuid); if (!conn) GOTO(out, rc = -ENOMEM); CDEBUG(D_HA, "switching import %s/%s from %s to %s\n", imp->imp_target_uuid.uuid, imp->imp_obd->obd_name, imp->imp_connection->c_remote_uuid.uuid, conn->c_remote_uuid.uuid); /* Switch the import's connection and the DLM export's * connection (which are almost certainly the same, but we * keep distinct refs just to make things clearer. I think. */ if (imp->imp_connection) ptlrpc_put_connection(imp->imp_connection); /* We hand off the ref from ptlrpc_get_connection. */ imp->imp_connection = conn; dlmexp = class_conn2export(&imp->imp_dlm_handle); if (dlmexp->exp_connection) ptlrpc_put_connection(dlmexp->exp_connection); dlmexp->exp_connection = ptlrpc_connection_addref(conn); class_export_put(dlmexp); } connect: rc = ptlrpc_connect_import(imp); if (rc < 0) { CERROR("failed to reconnect to %s@%s: %d\n", imp->imp_target_uuid.uuid, imp->imp_connection->c_remote_uuid.uuid, rc); GOTO(out, rc); } if (imp->imp_state == LUSTRE_IMP_EVICTED) { CDEBUG(D_HA, "evicted from %s@%s; invalidating\n", imp->imp_target_uuid.uuid, imp->imp_connection->c_remote_uuid.uuid); ptlrpc_set_import_active(imp, 0); CDEBUG(D_HA, "%s: new state: RECOVER\n", imp->imp_client->cli_name); imp->imp_state = LUSTRE_IMP_RECOVER; } if (imp->imp_state == LUSTRE_IMP_REPLAY) { CDEBUG(D_HA, "replay requested by %s\n", imp->imp_target_uuid.uuid); rc = ptlrpc_replay(imp); if (rc) GOTO(out, rc); rc = ldlm_replay_locks(imp); if (rc) GOTO(out, rc); rc = signal_completed_replay(imp); if (rc) GOTO(out, rc); CDEBUG(D_HA, "%s: new state: RECOVER\n", imp->imp_client->cli_name); imp->imp_state = LUSTRE_IMP_RECOVER; } if (imp->imp_state == LUSTRE_IMP_RECOVER) { CDEBUG(D_HA, "reconnected to %s@%s\n", imp->imp_target_uuid.uuid, imp->imp_connection->c_remote_uuid.uuid); ptlrpc_set_import_active(imp, 1); ptlrpc_resend(imp); spin_lock_irqsave(&imp->imp_lock, flags); CDEBUG(D_HA, "%s: new state: FULL\n", imp->imp_client->cli_name); imp->imp_state = LUSTRE_IMP_FULL; spin_unlock_irqrestore(&imp->imp_lock, flags); ptlrpc_wake_delayed(imp); } LASSERT(imp->imp_state == LUSTRE_IMP_FULL); out: if (rc != 0) { spin_lock_irqsave(&imp->imp_lock, flags); imp->imp_state = LUSTRE_IMP_DISCON; spin_unlock_irqrestore(&imp->imp_lock, flags); if (rc == -ENOTCONN) { CDEBUG(D_HA, "evicted/aborted by %s@%s during recovery;" "invalidating and reconnecting\n", imp->imp_target_uuid.uuid, imp->imp_connection->c_remote_uuid.uuid); GOTO(connect, -ENOTCONN); } else if (was_invalid) { ptlrpc_set_import_active(imp, 0); } } up(&imp->imp_recovery_sem); RETURN(rc); } void ptlrpc_fail_export(struct obd_export *exp) { int rc, already_failed; unsigned long flags; spin_lock_irqsave(&exp->exp_lock, flags); already_failed = exp->exp_failed; exp->exp_failed = 1; spin_unlock_irqrestore(&exp->exp_lock, flags); if (already_failed) { CDEBUG(D_HA, "disconnecting dead export %p/%s; skipping\n", exp, exp->exp_client_uuid.uuid); return; } CDEBUG(D_HA, "disconnecting export %p/%s\n", exp, exp->exp_client_uuid.uuid); /* Most callers into obd_disconnect are removing their own reference * (request, for example) in addition to the one from the hash table. * We don't have such a reference here, so make one. */ class_export_get(exp); rc = obd_disconnect(exp, 0); if (rc) CERROR("disconnecting export %p failed: %d\n", exp, rc); }