bacula crashes while starting backup

[Thomas Schweikle]

Jan 3 16:43:52 cbs1204 bacula-fd: cbs1204-fd: ABORTING due to ERROR in smartall.c:216#012Overrun buffer: len=9830441 addr=7f9d07710038 allocated: htable.c:72 called from htable.c:91
Jan 3 16:43:52 cbs1204 bacula-fd: Bacula interrupted by signal 11: Segmentation violation
Jan 3 16:51:02 cbs1204 bacula-fd: cbs1204-fd: ABORTING due to ERROR in smartall.c:216#012Overrun buffer: len=9830441 addr=7f9be171b038 allocated: htable.c:72 called from htable.c:91
Jan 3 16:51:02 cbs1204 bacula-fd: Bacula interrupted by signal 11: Segmentation violation
Jan 3 01:21:11 cbs1204 bacula-fd: cbs1204-fd: ABORTING due to ERROR in smartall.c:216#012Overrun buffer: len=9830441 addr=17a09a8 allocated: htable.c:72 called from htable.c:91
Jan 3 01:21:11 cbs1204 bacula-fd: Bacula interrupted by signal 11: Segmentation violation

ProblemType: Bug
DistroRelease: Ubuntu 12.04
Package: bacula-client 5.2.1-0ubuntu2
ProcVersionSignature: Ubuntu 3.2.0-7.13-virtual 3.2.0-rc7
Uname: Linux 3.2.0-7-virtual x86_64
ApportVersion: 1.90-0ubuntu1
Architecture: amd64
Date: Tue Jan 3 16:52:12 2012
InstallationMedia: Ubuntu-Server 10.04.1 LTS "Lucid Lynx" - Release amd64 (20100816.2)
PackageArchitecture: all
ProcEnviron:
 PATH=(custom, user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: bacula
UpgradeStatus: Upgraded to precise on 2011-12-13 (20 days ago)

[Thomas Schweikle (tps)]

Maybe a security problem, because of the reported buffer overrun and sigsegfault ...

[Scott Moser (smoser)]

Hi, could you try to recreate on 12.04? the bacula version you reported this on is 5.2.1-0ubuntu2, but the current version in the 12.04 archive is 5.2.5-0ubuntu6.

Also, more information about your config would be required to help someone recreate.

[Thomas Schweikle (tps)]

Since migration there are about 1-2 crashes per week. The server is completely down afterwards. Sometimes it helps to CTRL-ALT-ESC, sometimes you'll have to "pull the cord" (hard reset). It is all times the same: no memory available and killed processes therefore. Since the very same server runs without any problem until bacula-director is started, I assume it's a problem related to bacula-director crunching up all memory. I've tried to give some more memory to the server, but this only helps for short. Soon or later the server crashes again, because of "out of memory" conditions. Doesn't really matter how much memory the server has.

[Thomas Schweikle (tps)]

This keeps on. After some time bacula exhausts all memory, the server starting to kill processes. Sometimes this kills ssh, then the server isn't reachable remotely any more, sometimes this kills other essential processes forcing the server down.
At the moment bacula isn't usable any more and badly broken with ubuntu.
Since the same version of bacula runs without problems on an other system, not equipped with ubuntu 12.04 LTS (in this case: FreeBSD current with MySQL for the database). , I assume this an ubuntu specific problem. Unsure where exactly the problem is located: bacula, database (PostgreSQL), or other parts of the system. Most logs written are not helpful at all, because they are shortened after hard reboots.

[Launchpad Janitor (janitor)]

[Expired for bacula (Ubuntu) because there has been no activity for 60 days.]

[Thomas Schweikle (tps)]

Fixed in Ubuntu 12.04.3 at least.