Ian Collins stated:
< Sean McGrath - Sun Microsystems Ireland wrote:
< 
< >Ian Collins stated:
< >< I can't find this in the bug database,  but I get the occasional lockup
< >< (uses 100% of a CPU) with ypserv on my NIS master, currently running
< >< build 36.
< >< 
< >< I have to pkill it to get name services back.  Is this a known bug, if
< >< not is there anything I can do to get some useful information next time
< >< it happens?
< >
< > I've seen this when a nis client goes haywire and polls the ypserv
< > continuously.  In my case, finding what network interface was having the 
most
< > traffic and then snooping that interface found the client and fixed the 
client.
< >  
< >  
< >
< Interesting, did restarting ypserv fix the problem, or did the client
< continue?

  'fixing' the client in this case ment restarting its jumpstart install if
  I recall properly.  Didn't have to go near the server side, ypserv calmed
  down after the client was rebooted.

< 
< I'll look next time.  In my case, it appears to be triggered when a zone
< serving a web site is being heavily used.  The pages it is serving are
< mounted from the global zone, so the is quite a bit of NFS traffic
< between the zones.  Problem is, I can't snoop inter-zone packets.

  Can dtrace like the following help ?
http://daemons.net/~matty/code/nfstrace
http://blogs.sun.com/roller/page/samf?entry=a_dtrace_provider_for_nfs


< 
< Ian

-- 
Sean.
.
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to