[Date Prev][Date Next]
Re: Possible bug? (again)
On Tue, 18 Jul 2000, Kurt D. Zeilenga wrote:
> You might try configuring with --with-yielding-select=no ...
That seems to have helped a bit. Truss still shows many time() calls in a
row, but it seems to be "doing" more in between and the CPU load is lower.
It would appear that I also underestimated the amount of queries that
running nss_ldap on the same box as Samba would generate, so the load
(now) appears to be less of a slapd bug and more actual work being done.
You'll recall that before, even when idle, slapd seemed to race with no
apparent errors, so things have definitely improved :-)
I also took Randy's advice and snagged a gdb package (thanks!)
I'd like to thank both Randy and Kurt for their help (and patience
:) through this...