2009/12/14 Jeff Ross <jr...@openvistas.net>: > Hi all, > > While doing some pgbench runs on a new server before I put in on-line, I > triggered a malloc: out of space in kmem_map panic. > > trace and ps (long) below, dmesg below that. > > I have adjusted sysctl values like so for postgres: > > # For PostgreSQL Port > kern.seminfo.semmni=1024 > kern.seminfo.semmns=9082 > kern.shminfo.shmall=128000 > kern.shminfo.shmmax=2028000000 > > I see softdep mentioned in the trace below, so here's /etc/fstab
"Doctor doctor.. It hurts when I do this.." "Well.. Don't do that!" Your problem is that the kernel has run out of kernel memory. Those knobs you all cranked up to eleventy billion consume kvm. The reason they are set to lower limits is to prevent the sort of situation you have encountered. When you crank them to eleventy billion, and then start eleventy billion processes that consume such resources. expect the possibility of issues.
