Hi, Thanks for this information. I look on backtrace with gdb and I have got this one. I hope this is what your except. I join more detail, could you read it? Thanks, Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 1132489024 (LWP 11645)] 0x0000003c024082e9 in pthread_mutex_lock () from /lib64/libpthread.so.0 (gdb) backtrace full #0 0x0000003c024082e9 in pthread_mutex_lock () from /lib64/libpthread.so.0 No symbol table info available. #1 0x00000000004c7ea1 in hdb_cache_delete_cleanup () No symbol table info available. #2 0x00000000004c897c in hdb_cache_modrdn () No symbol table info available. #3 0x00000000004c9384 in hdb_cache_find_id () No symbol table info available. #4 0x00000000004cd193 in hdb_dn2entry () No symbol table info available. #5 0x00000000004a6513 in hdb_search () No symbol table info available. #6 0x00000000004314b6 in fe_op_search () No symbol table info available. #7 0x0000000000431c47 in do_search () No symbol table info available. #8 0x000000000042fa5b in connection_hangup () No symbol table info available. #9 0x0000000000430157 in connection_hangup () No symbol table info available. #10 0x000000000051f15a in ldap_pvt_thread_pool_submit () No symbol table info available. #11 0x0000003c024062e7 in start_thread () from /lib64/libpthread.so.0 No symbol table info available. #12 0x0000003c018ce3bd in clone () from /lib64/libc.so.6 No symbol table info available. -----Original Message----- From: Quanah Gibson-Mount [mailto:quanah@zimbra.com] Sent: Wednesday, April 22, 2009 7:25 PM To: Luca Scamoni Cc: SAGNIMORTE Thomas (CAMPUS); openldap-technical@openldap.org Subject: Re: Segfault during Auth Benchmark request --On Wednesday, April 22, 2009 7:23 PM +0200 Luca Scamoni <luca.scamoni@sys-net.it> wrote: > Quanah Gibson-Mount ha scritto: >> --On Wednesday, April 22, 2009 10:52 AM +0200 "SAGNIMORTE Thomas >> >>> I set entry cache to 100000, idlcache to 10000 and dncache to 10000. >> >> Seems reasonable. > > Since cachesize is 100K entries shouldn't idlcachesize be 300K ? (he's > using hdb backend) Hm, I read that all as 10k. I'm guessing that a 100k cachesize and a 10k idlcachesize wouldn't be greatly productive, but I've often run my HDB backends with the same cachesize/idlcachesize without ill effect. --Quanah -- Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
Attachment:
gdb-slapd2.txt.zip
Description: =?US-ASCII?Q?gdb-slapd2=2Etxt_Compresse?=