[ClusterLabs] Cannot allocate memory in pgsql_monitor

[Ларионов Андрей Валентинович]

Hello,

We have PostgreSQL under pacemaker/corosync management
and sometimes we catch problem like this:
Dec  9 18:51:15 prod-inside-tvh-pgsql1 lrmd[13405]:  notice: pgsql_monitor_8000:34884:stderr [ /usr/lib/ocf/lib/heartbeat/ocf-shellfuncs: fork: Cannot allocate memory ]
Dec  9 18:51:15 prod-inside-tvh-pgsql1 crmd[13409]:  notice: prod-inside-tvh-pgsql1-pgsql_monitor_8000:13 [ /usr/lib/ocf/lib/heartbeat/ocf-shellfuncs: fork: Cannot allocate memory\n ]
Dec  9 18:51:15 prod-inside-tvh-pgsql1 crmd[13409]:  notice: Result of notify operation for pgsql on prod-inside-tvh-pgsql1: 0 (ok)
Dec  9 18:51:15 prod-inside-tvh-pgsql1 crmd[13409]:  notice: prod-inside-tvh-pgsql1-pgsql_monitor_8000:13 [ /usr/lib/ocf/lib/heartbeat/ocf-shellfuncs: fork: Cannot allocate memory\n ]
Dec  9 18:51:15 prod-inside-tvh-pgsql1 pgsql(pgsql)[35130]: INFO: stop_escalate(or stop_escalate_in_slave) time is adjusted to 50 based on the configured timeout.
Dec  9 18:51:15 prod-inside-tvh-pgsql1 pgsql(pgsql)[35130]: INFO: server shutting down
Dec  9 18:51:17 prod-inside-tvh-pgsql1 pgsql(pgsql)[35130]: INFO: PostgreSQL is down
Dec  9 18:51:17 prod-inside-tvh-pgsql1 pgsql(pgsql)[35130]: INFO: Changing pgsql-status on prod-inside-tvh-pgsql1 : HS:async->STOP.
Dec  9 18:51:17 prod-inside-tvh-pgsql1 crmd[13409]:  notice: Result of stop operation for pgsql on prod-inside-tvh-pgsql1: 0 (ok)

Main - "/usr/lib/ocf/lib/heartbeat/ocf-shellfuncs: fork: Cannot allocate memory".
But memory is always enough.
This is virtual machine with Linux 4.14.35-1818.4.7.el7uek.x86_64
and
pacemaker 1.1.19-8.el7_6.4
corosync 2.4.3-4.el7
(all from linux repository).

Please, help us to resolve this problem with "Cannot allocate memory".
May be we need some additional settings on OS or upgrade to fresh pacemaker/corosync version where this issue resolved or some other?


--
WBR,
Andrey Larionov

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.clusterlabs.org/pipermail/users/attachments/20201210/cbc3e900/attachment.htm>