From: Donald H Locker (dhlocker@comcast.net)
Date: Sat Mar 29 2008 - 07:57:17 EST
Solaris 10, Sun Ultra 20 M2 (AMD processor); most recommended patches installed
until mid-2007.
Problem: several times a day, rpcbind will begin consuming all of one
processor's CPU time (ps shows it running at 49.3 to 49.8% CPU) when normally it
just uses 0.0% and sits there.)
Looks like statd and lockd are making unusual demands on rpcbind.
/etc/svc/volatile/{nfs-nlockmgr.lock,nfs-status.lock} have date/time that
coincide with rpcbind getting bound-up. netstat shows many bindings of sunrpc
with localhost.non service-specific ports in a CLOSE-WAIT state.
Local Address Remote Address Swind Send-Q Rwind Recv-Q State
localhost.sunrpc localhost.626 49152 0 49152 0 CLOSE_WAIT
svcadm restart any of statd, lockd, rstat have no effect, though one of those
appears to be the culprit.
pkill rpcbind restores sanity, but that's a sledgehammer approach, I think.
Suggestions of how to further the diagnosis, prevent the problem (patch or hack
welcome) or even sympathy welcome.
TIA,
Donald.
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:42:53 EDT