2013/7/8 Claudio Jeker <[email protected]> > > This is not netstat -m output. All the important information got stripped.
.. > This is as useless as the mangled netstat output. > .. > 508 connections, question is what kind of connections? > If it happens again I'll get this info then. > > - vmstat -m shows: > > 5842/5938/6144 mbuf 2048 byte clusters in use (current/peak/max) > > In use 126920K, total allocated: 158580K,; utlilization 80% > > and all is 0 in the fail column. > > You are running very close to the limit of mclusters. Where did all those > cluster go? Maybe in the 508 connections reported by netstat. > netstat show the buffer space used by the various sockets, maybe check > those. Once you figured out where the memory is gone to and why, you > should be able to figure out if it is save to increase nmbcluster because > your server needs them or if the problem is a leak of some sort. > What would be OpenBSD's ordinary transient and lasting behavior if mclusters run out? (I guess mclusters are supergroups of mbufs?) So, my nmbcluster is set to 200MB so just in case there was a peak mcluster use of 200MB, there would be allocation failures? Do you have any thought on where the mclusters may have went? Now - at the time netstat dumped that output, utilization was indeed stable at 80%, and even then the "hangup error" was still effective. Does this make any sense to you? Thanks, Mikael
