Error in Directory server - suggestions..
Hi..
We are running a 5.2 Directory server + Mesaging server on a Sun Solaris server. This weekend our disks grew full and the Directory indexing job crashed. After this incident th directory server isn磘 workin the way it should. Everything starts up ok and runs for an hour, then the indexing job starts and crashes the slapd process - stops responding.
The following entry is found n the errorlog:
[04/Jul/2006:23:42:10 +0200] - ERROR<20775> - Backend Database - conn=6747 op=1 msgId=2 - database error -30989
[04/Jul/2006:23:42:40 +0200] - DEBUG - conn=-1 op=-1 msgId=-1 - database index operation failed BAD 1050, err=-30989 DB_RUNRECOVERY: Fatal error, run database recovery
[04/Jul/2006:23:42:40 +0200] - ERROR<20775> - Backend Database - conn=6748 op=1 msgId=2 - database error -30989
[04/Jul/2006:23:42:52 +0200] - ERROR<20504> - Backend Database - conn=-1 op=-1 msgId=-1 - Serious failure in deadlock detect, err=-30989 (DB_RUNRECOVERY: Fatal error, run database recovery)
[04/Jul/2006:23:42:56 +0200] - ERROR<20506> - Backend Database - conn=-1 op=-1 msgId=-1 - Serious failure during trickle, err=-30989 (DB_RUNRECOVERY: Fatal error, run database recovery)
[04/Jul/2006:23:43:10 +0200] - DEBUG - conn=-1 op=-1 msgId=-1 - database index operation failed BAD 1050, err=-30989 DB_RUNRECOVERY: Fatal error, run database recovery
[04/Jul/2006:23:43:10 +0200] - ERROR<20775> - Backend Database - conn=6749 op=1 msgId=2 - database error -30989
After stopping messaging and directory server, then starting slapd-server the following appears in the log:
[04/Jul/2006:23:45:22 +0200] - DEBUG - conn=-1 op=-1 msgId=-1 - libdb: region error detected; run recovery.
Another stop and start eliminates that message and everything seems to working for an hour until the next dbindex is run.
I have recreated the indexes with db2index and everything seems to work ok with that utility. The indexes are recreated with no error. But the error remains...
Any suggestions? Could it be the transaction log that is corrupt in some way..?
//Bns
