Steven Chamberlain wrote: > kernel: maxproc limit exceeded by uid 0, please see tuning(7) and login.conf(5). I only ever saw this happen on one server, which I no longer have, and could never reproduce it anywhere else. nagios3/icinga probes for a running SSH service could trigger it after some time. A public-facing SSH service could get into this state even more quickly due to frequent connections by botnets. Killing some random process on the system would sometimes recover it from this state, until it happens again. It never seemed that an excessive number of processes was actually running, so I thought it could be a bug in the kernel's counting of processes or threads, or maybe in glibc's reclaimation of resources after threads exit. Regards, -- Steven Chamberlain steven@pyro.eu.org
Attachment:
signature.asc
Description: Digital signature