Quanah and Howard, Sorry to come back.
I tried to make 2 ldapsearch to the same DB simultaneously, doing some stress test like dump the LDIF and queries coming. The slapd after sometime start to get more memory. By what I could observe looks like new thread was started consuming again the same memory as before so the slad now is consuming around the double of memory as before. I believe the cache would not grow beyond the boundary as like per thread since once cache is allocated it is never released. PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 10195 ldap 18 0 1913m 1.7g 68m S 100 15.0 31:25.71 slapd Then performance start to degrade again. Like before looks like the dump get stuck and comes in chunks. See example below : Mon Jan 26 21:05:16 BRST 2009 378752 Mon Jan 26 21:09:21 BRST 2009 378800 See that after around 4 minutes I have only 48 adds to the output LDIF from my first ldapsearch. Looks like if multiples queries are happening something still getting lost. Both queries are to the same DB and the monitor has : readOnly: FALSE olmBDBEntryCache: 1000 olmBDBDNCache: 1000953 olmBDBIDLCache: 1 Sorry to come back with this issue but I could detect this now. Thanks, Rodrigo. --- On Mon, 1/26/09, Quanah Gibson-Mount <[email protected]> wrote: > From: Quanah Gibson-Mount <[email protected]> > Subject: Re: (ITS#5860) slapd memeory leak under openldap 2.4 > To: [email protected], [email protected] > Date: Monday, January 26, 2009, 8:53 PM > --On Monday, January 26, 2009 10:44 PM +0000 > [email protected] wrote: > > > Just some last questions. Is there any expectation for > a new release, for > > example 2.4.15, with this ITS included? Just to use a > formal release. > > 2.4.14 will have the fix for this ITS. There's no set > release date at this time, but hopefully soon. > > --Quanah > > -- > > Quanah Gibson-Mount > Principal Software Engineer > Zimbra, Inc > -------------------- > Zimbra :: the leader in open source messaging and > collaboration
