Hi fellow administrators:
THE PROBLEM:
I am experiencing the following problem with NFS on an Ultra Sparc
Model 140 running Solaris 2.5.1 (OS 5.5.1):
Feb 28 09:27:25 ew06 automountd[151]: server ew01 not responding
Feb 28 09:27:40 ew06 automountd[151]: server ew05 not responding
Both of the servers above are Sparc 2's running OS 4.1.3 (slow servers).
INDICATIONS:
The true nature of the problem was best indicated by running
vmstat -s on the Ultra Sparc client. Near the end of this report:
20019064 total name lookups (cache hits 56%)
The cache hits were too low. Acceptable performance is only achieved
when cache hits are > 90%.
THE SOLUTION:
The largest single improvement in performance is due to a single change
on the Ultra Sparc Client. Adding the actimeo=300 option to its
/etc/auto.master file for all server mount points (including those above) as
follows made a LARGE difference in observable performance. Cache hits now
average 98%.
CREDIT:
(See References)
No one responded to this query. It was posted two months ago.
References:
1) "SMCC NFS Performance and Tuning Guide", Solaris 2.5 SMCC Manual,
Sun Microsystems. (Available in the on-line answerbook).
2) Cockcroft, Adrian. "Sun Performance and Tuning", Sun Microsystems,
1995.
3) Stern, Hal. "Managing NFS and NIS", O'Reilly.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:50 CDT