[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: An interesting PANIC



Frank Swasey wrote:

OpenLDAP: 2.2.24
OS: Red Hat Enterprise Linux AS release 3 (Taroon Update 4)

Problem: bdb(dc=uvm,dc=edu): PANIC: fatal region error detected; run recovery

Somehow a bdb log file got created and owned by root. Based on the timestamp and the contents of the log, it appears that it happened because the nightly slapcat (which runs as root) did a checkpoint. Either that or it was absolute coincidence that the checkpoint was run at the exact same time (00:17:09) as the slapcat. Slapd runs as user ldap, so I don't know how slapd would have created the log file owned by root.

Should slapcat be doing checkpoints? Is this a known problem? Do I need to change the way the ldap user is defined on the Red Hat system so the slapcat can run as user ldap?

slapcat invokes the usual backend open and close routines, and the back-bdb close routine always does a checkpoint before closing its environment handle. Perhaps it should skip this call when accessing the database in read-only mode.


--
 -- Howard Chu
 Chief Architect, Symas Corp.       Director, Highland Sun
 http://www.symas.com               http://highlandsun.com/hyc
 Symas: Premier OpenSource Development and Support