Re: What causes this error?
At Mon, 15 Nov 2010 18:03:23 -0600 Rhys Rhaven r...@rhavenindustrys.com wrote: Looks like the bug that I posted, which I believe Jean-Louis sent the patch for. I've been unable to test it as of yet. I compared my logs with your logs and it is much the same. I cannot find the patch you mentioned in the E-Mail archives. Can you (or Jean-Louis) re-post it? On 11/15/2010 04:11 PM, Robert Heller wrote: At Mon, 15 Nov 2010 15:34:31 -0500 Jean-Louis Martineaumartin...@zmanda.com wrote: Because ravel:/mnt/RAID0 was not dumped. Check the email report to know why or the client debug file (on ravel) The E-Mail report just says: FAILURE DUMP SUMMARY: ravel /mnt/RAID0 RESULTS MISSING ravel /mnt/RAID0 lev 1 FAILED [can't dump in degraded mode] And there isn't a sendbackup debug file for /mnt/RAID0. There are sendbackup debug files for the other file systems on ravel. There is a sendsize debug file for /mnt/RAID0 on ravel. Jean-Louis Robert Heller wrote: Using /var/log/amanda/60villagedrive/amdump.1 From Mon Nov 15 00:10:02 EST 2010 bach:/ 183m finished (0:17:59) haydn:/ 147m finished (0:20:35) ravel:/ 189m finished (0:20:36) ravel:/mnt/RAID0 1 816m failed: process terminated while waiting for dumping ravel:/mnt/sdb2 1 0m finished (0:20:34) .. This is in a CentOS 5.5 system: -sh-3.2$ rpm -qa amanda\* perl dump amanda-backup_server-3.2.0-1.rhel5 dump-0.4b41-4.el5 perl-5.8.8-32.el5_5.2 -- Robert Heller -- 978-544-6933 / hel...@deepsoft.com Deepwoods Software-- http://www.deepsoft.com/ () ascii ribbon campaign -- against html e-mail /\ www.asciiribbon.org -- against proprietary attachments
Re: What causes this error?
I attach the patch that was committed. Jean-Louis Robert Heller wrote: At Mon, 15 Nov 2010 18:03:23 -0600 Rhys Rhaven r...@rhavenindustrys.com wrote: Looks like the bug that I posted, which I believe Jean-Louis sent the patch for. I've been unable to test it as of yet. I compared my logs with your logs and it is much the same. I cannot find the patch you mentioned in the E-Mail archives. Can you (or Jean-Louis) re-post it? Index: server-src/driver.c === --- server-src/driver.c (revision 3607) +++ server-src/driver.c (revision 3608) @@ -499,32 +499,10 @@ headqueue_disk(directq, diskp); } -/* handle any remaining dumps by dumping directly to tape, if possible */ -while(!empty(directq) taper_fd 0) { - time_t sleep_time = 1; - disk_t *sleep_diskp = NULL; - time_t now = time(0); +/* log error for any remaining dumps */ +while(!empty(directq)) { + diskp = dequeue_disk(directq); - /* Find one we can do immediately or the sonner */ - for (diskp = directq.head; diskp != NULL; diskp = diskp-next) { - if (diskp-to_holdingdisk == HOLD_REQUIRED || - degraded_mode) { - sleep_time = 0; - sleep_diskp = diskp; - } else if (diskp-host-start_t - now sleep_time - diskp-start_t -now sleep_time) { - if (diskp-host-start_t diskp-start_t) - sleep_time = diskp-host-start_t - now; - else - sleep_time = diskp-start_t - now; - sleep_diskp = diskp; - } - } - diskp = sleep_diskp; - if (sleep_time 0) - sleep(sleep_time); - remove_disk(directq, diskp); - if (diskp-to_holdingdisk == HOLD_REQUIRED) { char *qname = quote_string(diskp-name); log_add(L_FAIL, %s %s %s %d [%s], @@ -538,7 +516,7 @@ log_add(L_FAIL, %s %s %s %d [%s], diskp-host-hostname, qname, sched(diskp)-datestamp, sched(diskp)-level, - _(can't dump in degraded mode)); + _(can't dump in non degraded mode)); amfree(qname); } else { @@ -555,10 +533,6 @@ } } -/* fill up the tape or start new one for taperflush */ -startaflush(); -event_loop(0); - short_dump_state();/* for amstatus */ g_printf(_(driver: QUITTING time %s telling children to quit\n), @@ -1044,6 +1018,7 @@ enqueue_disk(directq, diskp); diskp-to_holdingdisk = HOLD_NEVER; } + if (empty(*rq)) force_flush = 1; } } else if (client_constrained(diskp)) { free_assignedhd(holdp); @@ -1326,6 +1301,7 @@ dumper_cmd(dumper, PORT_DUMP, diskp, NULL); } diskp-host-start_t = now + 15; + if (empty(*rq)) force_flush = 1; if (result_argv) g_strfreev(result_argv); @@ -2114,6 +2090,7 @@ dp-inprogress = 0; deallocate_bandwidth(dp-host-netif, sched(dp)-est_kps); taper-dumper = NULL; +taper-disk = NULL; sched(dp)-dumper = NULL; sched(dp)-taper = NULL; start_some_dumps(runq); @@ -2759,7 +2736,7 @@ continue; } - if(file.dumplevel 0 || file.dumplevel 9) { + if (file.dumplevel 0 || file.dumplevel 399) { log_add(L_INFO, _(%s: ignoring file with bogus dump level %d.), destname, file.dumplevel); amfree(destname); @@ -2825,6 +2802,8 @@ if (!nodump) { schedule_ev_read = event_register((event_id_t)0, EV_READFD, read_schedule, NULL); +} else { + force_flush = 1; } } @@ -3137,6 +3116,7 @@ log_add(L_WARNING, _(WARNING: got empty schedule from planner)); if(need_degraded==1) start_degraded_mode(runq); schedule_done = 1; +if (empty(runq)) force_flush = 1; start_some_dumps(runq); startaflush(); } @@ -3805,13 +3785,9 @@ } // when to start a flush -// We don't start a flush if taper_tape_started == 1 dump_to_disk_terminated force_flush == 0, -// it is a criteria need to exit the first event_loop without flushing everything to tape, -// they will be flush in another event_loop. if (taper-state TAPER_STATE_IDLE) { if (!degraded_mode (!empty(tapeq) || !empty(directq)) - (((taper-state TAPER_STATE_TAPE_STARTED) - force_flush == 1) ||// if tape already started and force_flush + (taper-state TAPER_STATE_TAPE_STARTED || // tape already started !empty(roomq) || // holding disk constraint idle_reason == IDLE_NO_DISKSPACE || // holding disk constraint (my_flush_threshold_dumped tapeq_size // flush-threshold-dumped
What causes this error?
Using /var/log/amanda/60villagedrive/amdump.1 From Mon Nov 15 00:10:02 EST 2010 bach:/ 183m finished (0:17:59) haydn:/ 147m finished (0:20:35) ravel:/ 189m finished (0:20:36) ravel:/mnt/RAID0 1 816m failed: process terminated while waiting for dumping ravel:/mnt/sdb2 1 0m finished (0:20:34) .. This is in a CentOS 5.5 system: -sh-3.2$ rpm -qa amanda\* perl dump amanda-backup_server-3.2.0-1.rhel5 dump-0.4b41-4.el5 perl-5.8.8-32.el5_5.2 -- Robert Heller -- 978-544-6933 / hel...@deepsoft.com Deepwoods Software-- http://www.deepsoft.com/ () ascii ribbon campaign -- against html e-mail /\ www.asciiribbon.org -- against proprietary attachments
Re: What causes this error?
Because ravel:/mnt/RAID0 was not dumped. Check the email report to know why or the client debug file (on ravel) Jean-Louis Robert Heller wrote: Using /var/log/amanda/60villagedrive/amdump.1 From Mon Nov 15 00:10:02 EST 2010 bach:/ 183m finished (0:17:59) haydn:/ 147m finished (0:20:35) ravel:/ 189m finished (0:20:36) ravel:/mnt/RAID0 1 816m failed: process terminated while waiting for dumping ravel:/mnt/sdb2 1 0m finished (0:20:34) .. This is in a CentOS 5.5 system: -sh-3.2$ rpm -qa amanda\* perl dump amanda-backup_server-3.2.0-1.rhel5 dump-0.4b41-4.el5 perl-5.8.8-32.el5_5.2
Re: What causes this error?
At Mon, 15 Nov 2010 15:34:31 -0500 Jean-Louis Martineau martin...@zmanda.com wrote: Because ravel:/mnt/RAID0 was not dumped. Check the email report to know why or the client debug file (on ravel) The E-Mail report just says: FAILURE DUMP SUMMARY: ravel /mnt/RAID0 RESULTS MISSING ravel /mnt/RAID0 lev 1 FAILED [can't dump in degraded mode] And there isn't a sendbackup debug file for /mnt/RAID0. There are sendbackup debug files for the other file systems on ravel. There is a sendsize debug file for /mnt/RAID0 on ravel. Jean-Louis Robert Heller wrote: Using /var/log/amanda/60villagedrive/amdump.1 From Mon Nov 15 00:10:02 EST 2010 bach:/ 183m finished (0:17:59) haydn:/ 147m finished (0:20:35) ravel:/ 189m finished (0:20:36) ravel:/mnt/RAID0 1 816m failed: process terminated while waiting for dumping ravel:/mnt/sdb2 1 0m finished (0:20:34) .. This is in a CentOS 5.5 system: -sh-3.2$ rpm -qa amanda\* perl dump amanda-backup_server-3.2.0-1.rhel5 dump-0.4b41-4.el5 perl-5.8.8-32.el5_5.2 -- Robert Heller -- 978-544-6933 / hel...@deepsoft.com Deepwoods Software-- http://www.deepsoft.com/ () ascii ribbon campaign -- against html e-mail /\ www.asciiribbon.org -- against proprietary attachments
Re: What causes this error?
Looks like the bug that I posted, which I believe Jean-Louis sent the patch for. I've been unable to test it as of yet. On 11/15/2010 04:11 PM, Robert Heller wrote: At Mon, 15 Nov 2010 15:34:31 -0500 Jean-Louis Martineaumartin...@zmanda.com wrote: Because ravel:/mnt/RAID0 was not dumped. Check the email report to know why or the client debug file (on ravel) The E-Mail report just says: FAILURE DUMP SUMMARY: ravel /mnt/RAID0 RESULTS MISSING ravel /mnt/RAID0 lev 1 FAILED [can't dump in degraded mode] And there isn't a sendbackup debug file for /mnt/RAID0. There are sendbackup debug files for the other file systems on ravel. There is a sendsize debug file for /mnt/RAID0 on ravel. Jean-Louis Robert Heller wrote: Using /var/log/amanda/60villagedrive/amdump.1 From Mon Nov 15 00:10:02 EST 2010 bach:/ 183m finished (0:17:59) haydn:/ 147m finished (0:20:35) ravel:/ 189m finished (0:20:36) ravel:/mnt/RAID0 1 816m failed: process terminated while waiting for dumping ravel:/mnt/sdb2 1 0m finished (0:20:34) .. This is in a CentOS 5.5 system: -sh-3.2$ rpm -qa amanda\* perl dump amanda-backup_server-3.2.0-1.rhel5 dump-0.4b41-4.el5 perl-5.8.8-32.el5_5.2