2 Bacula® - The Network Backup Solution
4 Copyright (C) 2000-2007 Free Software Foundation Europe e.V.
6 The main author of Bacula is Kern Sibbald, with contributions from
7 many others, a complete list can be found in the file AUTHORS.
8 This program is Free Software; you can redistribute it and/or
9 modify it under the terms of version two of the GNU General Public
10 License as published by the Free Software Foundation plus additions
11 that are listed in the file LICENSE.
13 This program is distributed in the hope that it will be useful, but
14 WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 Bacula® is a registered trademark of John Walker.
24 The licensor of Bacula is the Free Software Foundation Europe
25 (FSFE), Fiduciary Program, Sumatrastrasse 25, 8006 Zürich,
26 Switzerland, email:ftf@fsfeurope.org.
29 * Subroutines to handle waiting for operator intervention
30 * or waiting for a Device to be released
32 * Code for wait_for_sysop() pulled from askdir.c
34 * Kern Sibbald, March 2005
40 #include "bacula.h" /* pull in global headers */
41 #include "stored.h" /* pull in Storage Deamon headers */
43 //static bool double_jcr_wait_time(JCR *jcr);
47 * Wait for SysOp to mount a tape on a specific device
49 * Returns: W_ERROR, W_TIMEOUT, W_POLL, W_MOUNT, or W_WAKE
51 int wait_for_sysop(DCR *dcr)
55 struct timespec timeout;
56 time_t last_heartbeat = 0;
57 time_t first_start = time(NULL);
61 DEVICE *dev = dcr->dev;
65 Dmsg1(100, "Enter blocked=%s\n", dev->print_blocked());
66 unmounted = is_device_unmounted(dev);
70 * Wait requested time (dev->rem_wait_sec). However, we also wake up every
71 * HB_TIME seconds and send a heartbeat to the FD and the Director
72 * to keep stateful firewalls from closing them down while waiting
75 add_wait = dev->rem_wait_sec;
76 if (me->heartbeat_interval && add_wait > me->heartbeat_interval) {
77 add_wait = me->heartbeat_interval;
79 /* If the user did not unmount the tape and we are polling, ensure
80 * that we poll at the correct interval.
82 if (!unmounted && dev->vol_poll_interval && add_wait > dev->vol_poll_interval) {
83 add_wait = dev->vol_poll_interval;
87 Dmsg1(400, "blocked=%s\n", dev->print_blocked());
88 dev->dev_prev_blocked = dev->blocked();
89 dev->set_blocked(BST_WAITING_FOR_SYSOP); /* indicate waiting for mount */
92 for ( ; !job_canceled(jcr); ) {
95 gettimeofday(&tv, &tz);
96 timeout.tv_nsec = tv.tv_usec * 1000;
97 timeout.tv_sec = tv.tv_sec + add_wait;
99 Dmsg4(400, "I'm going to sleep on device %s. HB=%d wait=%d add_wait=%d\n",
100 dev->print_name(), (int)me->heartbeat_interval, dev->wait_sec, add_wait);
102 /* Wait required time */
103 stat = pthread_cond_timedwait(&dev->wait_next_vol, &dev->m_mutex, &timeout);
104 Dmsg2(400, "Wokeup from sleep on device stat=%d blocked=%s\n", stat,
105 dev->print_blocked());
108 dev->rem_wait_sec -= (now - start);
110 /* Note, this always triggers the first time. We want that. */
111 if (me->heartbeat_interval) {
112 if (now - last_heartbeat >= me->heartbeat_interval) {
113 /* send heartbeats */
114 if (jcr->file_bsock) {
115 jcr->file_bsock->signal(BNET_HEARTBEAT);
116 Dmsg0(400, "Send heartbeat to FD.\n");
118 if (jcr->dir_bsock) {
119 jcr->dir_bsock->signal(BNET_HEARTBEAT);
121 last_heartbeat = now;
125 if (stat == EINVAL) {
127 Jmsg1(jcr, M_FATAL, 0, _("pthread timedwait error. ERR=%s\n"), be.bstrerror(stat));
128 stat = W_ERROR; /* error */
133 if (dev->rem_wait_sec <= 0) { /* on exceeding wait time return */
134 Dmsg0(400, "Exceed wait time.\n");
140 * Check if user unmounted the device while we were waiting
142 unmounted = is_device_unmounted(dev);
144 if (!unmounted && dev->vol_poll_interval &&
145 (now - first_start >= dev->vol_poll_interval)) {
146 Dmsg1(400, "In wait blocked=%s\n", dev->print_blocked());
147 dev->poll = true; /* returning a poll event */
152 * Check if user mounted the device while we were waiting
154 if (dev->blocked() == BST_MOUNT) { /* mount request ? */
160 * If we did not timeout, then some event happened, so
161 * return to check if state changed.
164 stat = W_WAKE; /* someone woke us */
169 * At this point, we know we woke up because of a timeout,
170 * that was due to a heartbeat, so we just update
171 * the wait counters and continue.
173 add_wait = dev->wait_sec - (now - start);
177 if (me->heartbeat_interval && add_wait > me->heartbeat_interval) {
178 add_wait = me->heartbeat_interval;
183 dev->set_blocked(dev->dev_prev_blocked); /* restore entry state */
184 Dmsg1(400, "set %s\n", dev->print_blocked());
186 Dmsg1(400, "Exit blocked=%s\n", dev->print_blocked());
193 * Wait for any device to be released, then we return, so
194 * higher level code can rescan possible devices. Since there
195 * could be a job waiting for a drive to free up, we wait a maximum
196 * of 1 minute then retry just in case a broadcast was lost, and
197 * we return to rescan the devices.
199 * Returns: true if a device has changed state
200 * false if the total wait time has expired.
202 bool wait_for_device(JCR *jcr, int &retries)
206 struct timespec timeout;
209 const int max_wait_time = 1 * 60; /* wait 1 minute */
212 Dmsg0(100, "Enter wait_for_device\n");
213 P(device_release_mutex);
215 if (++retries % 5 == 0) {
216 /* Print message every 5 minutes */
217 Jmsg(jcr, M_MOUNT, 0, _("JobId=%s, Job %s waiting to reserve a device.\n"),
218 edit_uint64(jcr->JobId, ed1), jcr->Job);
221 gettimeofday(&tv, &tz);
222 timeout.tv_nsec = tv.tv_usec * 1000;
223 timeout.tv_sec = tv.tv_sec + max_wait_time;
225 Dmsg1(100, "JobId=%u going to wait for a device.\n", (uint32_t)jcr->JobId);
227 /* Wait required time */
228 stat = pthread_cond_timedwait(&wait_device_release, &device_release_mutex, &timeout);
229 Dmsg2(100, "JobId=%u wokeup from sleep on device stat=%d\n", (uint32_t)jcr->JobId, stat);
231 V(device_release_mutex);
232 Dmsg2(100, "JobId=%u return from wait_device ok=%d\n", (uint32_t)jcr->JobId, ok);
238 * The jcr timers are used for waiting on any device *
239 * Returns: true if time doubled
240 * false if max time expired
242 static bool double_jcr_wait_time(JCR *jcr)
244 jcr->wait_sec *= 2; /* double wait time */
245 if (jcr->wait_sec > jcr->max_wait) { /* but not longer than maxtime */
246 jcr->wait_sec = jcr->max_wait;
249 jcr->rem_wait_sec = jcr->wait_sec;
250 if (jcr->num_wait >= jcr->max_num_wait) {