From: Robert Milkowski (rmilkowski@wp-sa.pl)
Date: Mon Oct 06 2003 - 06:26:19 EDT
Hi.
I've got server with Solaris 9 SPARC as nfs client. Under heavy IO
to nfs after some time (sometimes it's just after 15-30 minutes, sometimes
after 1-3 days) I get all processes using nfs filesystems stop working and
get 100% sys time.
Using lockstat I get:
Adaptive mutex spin: 16413 events in 10.015 seconds (1639 events/sec)
Count indv cuml rcnt spin Lock Caller
-------------------------------------------------------------------------------
7632 46% 46% 1.00 2 strresources getiocseqno+0xc
77 0% 47% 1.00 29 0x30003fcdec8 udp_bind+0x500
60 0% 47% 1.00 12 0x30003fcd908 udp_bind+0x500
33 0% 48% 1.00 6 0x30003fd1cc8 udp_bind+0x500
32 0% 48% 1.00 5 0x30003fd08c8 udp_bind+0x500
32 0% 48% 1.00 4 0x30003fd1b48 udp_bind+0x500
31 0% 48% 1.00 3 0x30003fcee88 udp_bind+0x500
I can do "ls" to nfs filesystems or "touch file". I can mount the same
filesystems on another server and start application - so it's not a
problem with nfs server.
The same application under Solaris 8 is working without any problems.
Any ideas? I can't find any information on sunsolve, etc. about
strresources or getiocseqno.
btw: Solaris 9 8/03 Sparc + lates patches. I got the same problem with
older Solaris 9 releases.
NFS v3 UDP
--
Robert Milkowski
rmilkowski@wp-sa.pl
_______________________________________________
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 - 23:27:14 EDT