Hi Tom:
On Tue, 14 Feb 2023 at 15:47, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> If overcommit-disabled doesn't seem practical, another idea that's
> been recommended is to start the postmaster under a "ulimit -v"
> setting chosen to cause plain ENOMEM failures before the OOM
> killer kicks in.
This may certainly help, but if I read the oomkiller stuff correctly
you can have a case where Pg has its (big) share of memory allocated
and commited, some other processes have overallocated and have
uncommitted memory and when one of these tries to commit some mem the
OOM kills Pg because it is the biggest memory user. Not an covercommti
fan, so not too sure.
FOS
