[Date Prev][Date Next]
Re: slapd crash: ch_malloc.c:77: failed assertion `0'
I believe folks have previously complained (see archives)
about leaks in Solaris threads and regex libraries. You might
verify you are running with latest versions from Sun.
At 08:18 AM 2001-12-08, Erik Persson wrote:
>On Sat, 8 Dec 2001, Erik Persson wrote:
>> I am running OpenLDAP 2.0.17 (I am going to get 2.0.18) with the ldbm
>> backend using Berkeley DB 3.3.11. After actualling having gone production
>> with the server (no major issues occured during the test phase that
>> couldn't be fixed) I now find that slapd dies sproradically but
>> frequently, usually with:
>> ch_malloc.c:77: failed assertion `0'
>> Could this be a symptom of the memory being exhausted?
>I did a little more research and I wonder if there might not be some kind
>of memory leak here. I have less that thousand dn:s in my LDAP namespace
>and the cache settings in slapd.conf have been reset to default. Still,
>when a hung application on my workstation caused nss_ldap to _hammer_ the
>slapd with queries, slapd grew from its initial 4 MB to more that 600 MB
>VSZ within five minutes. I am running OpenLDAP on a Sun Ultra 5 running
>a well-patched Solaris 8 and the software is in this case compiled with
>gcc (probably 2.95.3 but it is also possible that 3.0.2 was used).
>Erik Persson, System Manager <email@example.com>
>Roxen Internet Software Voice: +46 13 376817