The load average on the box in question almost never gets above 0.5.
The current load average is 0.01.
Owen
The fact that you are deleting the BDB environment files (The __* stuff) suggests some type of corruption. db_recover will correctly close those files out (and remove them). Deleting them manually is not a wise path to take.
Right... In my case, at least, the files become corrupt, ldap hangs on the corrupt files, slapd_db_recover will recover from the corruption, then, slaptest recreates these files with incorrect ownership, I fix the ownership, and start slapd, and things are back to normal until next corruption.
The bigger question in my mind is why were things getting corrupted in the first place. However, since implementing the ldlcache recommended and upgrading to 2.2.24 (I don't know which of these actions resolved the issue), I've run much longer than ever before without things freezing (about 3 days). We'll see how long it lasts.
How high is the load average on your box now?
Sam
Attachment:
pgpYuNT1Msu17.pgp
Description: PGP signature