Thread: pg_dump throwing segfault error during backup
Hello Team,
We are facing a segfault issue while initiating backup with pg_dump.
This is for Postgresql community version 12.5 and rhel 7.4 (PostgreSQL 12.5 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-39), 64-bit)
The below error is showing :
from /var/log/message
[root@USMS-PV-CRMDB2 log]# grep 'segfault' messages*
messages:Sep 24 00:15:56 USMS-PV-CRMDB2 kernel: postgres[1535]: segfault at 23c0000 ip 00007f272c91f61d sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages:Sep 25 00:14:59 USMS-PV-CRMDB2 kernel: postgres[4245]: segfault at 23c0000 ip 00007f272c91f61d sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages-20240923:Sep 21 00:07:13 USMS-PV-CRMDB2 kernel: postgres[14743]: segfault at 23c0000 ip 00007f272c91f622 sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages-20240923:Sep 22 00:06:53 USMS-PV-CRMDB2 kernel: postgres[392]: segfault at 23c0000 ip 00007f272c91f61d sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages-20240923:Sep 23 00:16:07 USMS-PV-CRMDB2 kernel: postgres[28843]: segfault at 23c0000 ip 00007f272c91f622 sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages:Sep 24 00:15:56 USMS-PV-CRMDB2 kernel: postgres[1535]: segfault at 23c0000 ip 00007f272c91f61d sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages:Sep 25 00:14:59 USMS-PV-CRMDB2 kernel: postgres[4245]: segfault at 23c0000 ip 00007f272c91f61d sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages-20240923:Sep 21 00:07:13 USMS-PV-CRMDB2 kernel: postgres[14743]: segfault at 23c0000 ip 00007f272c91f622 sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages-20240923:Sep 22 00:06:53 USMS-PV-CRMDB2 kernel: postgres[392]: segfault at 23c0000 ip 00007f272c91f61d sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
messages-20240923:Sep 23 00:16:07 USMS-PV-CRMDB2 kernel: postgres[28843]: segfault at 23c0000 ip 00007f272c91f622 sp 00007ffe22636718 error 6 in libc-2.17.so[7f272c7d3000+1b8000]
From db log :
2024-09-26 00:47:05.312 +08 [9908] HINT: In a moment you should be able to reconnect to the database and repeat your command.
2024-09-26 00:47:05.306 +08 [10870] WARNING: terminating connection because of crash of another server process
2024-09-26 00:47:05.306 +08 [10870] DETAIL: The postmaster has commanded this server process to roll back the current transaction and exit, because another server process exited abnormally and possibly corrupted shared memory.
2024-09-26 00:47:05.306 +08 [10870] WARNING: terminating connection because of crash of another server process
2024-09-26 00:47:05.306 +08 [10870] DETAIL: The postmaster has commanded this server process to roll back the current transaction and exit, because another server process exited abnormally and possibly corrupted shared memory.
We fond the timing which shows database backup is getting invoked at the time. So we had to stop backup (pg_backup)
we are manually taking table data currently.
Core dump also generated however RedHat refused to analyze this saying it is third party product.
Kindly let us know if this is a bug for the version as it is too old and if the core dump can be analyzed somehow as it is in binary.
Regards
Somraj Chakrabarty
Cassandra Thrift <cssndrthrift@gmail.com> writes:
> We are facing a segfault issue while initiating backup with pg_dump.
> This is for Postgresql community version 12.5 and rhel 7.4 (PostgreSQL 12.5
> on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623 (Red Hat
> 4.8.5-39), 64-bit)
The current release of PG12 is 12.20, so you are missing fifteen minor
releases' worth of bug fixes, one of which might well have addressed
your problem. You should update to 12.20 before spending any further
effort on investigation.
Also ... PG12 will be out of support altogether after next month's
12.21 release. And I'm astonished that Red Hat would provide any
support whatsoever for RHEL7.4, which is about seven years obsolete
as well. You really need to make plans to migrate onto supported
versions sooner not later.
regards, tom lane