Yesterday, I upgraded our 6.1.4.1 server to 6.1.4.3. All hell broke loose last night and the server crashed-and-burned (called at 2am).
Unfortunately, the crash seems to have wiped a lot of the activity log for the time period when the failures started to occur (q actlog doesn't show anything beyond 19:44 yesterday) Fortunately, the TSMManager console captured a lot of the error messages, which look like this: 11/8/2010 9:54:42 PM ANR0101E bfdedup.c(8813): Error 11 opening table "BF.Dereferenced.Chunks". 11/8/2010 9:54:53 PM ANR0101E bfdedup.c(8813): Error 11 opening table "BF.Dereferenced.Chunks". 11/8/2010 9:55:04 PM ANR0101E bfdedup.c(8813): Error 11 opening table "BF.Dereferenced.Chunks". 11/8/2010 9:55:16 PM ANR0101E bfdedup.c(8813): Error 11 opening table "BF.Dereferenced.Chunks". 11/8/2010 9:55:27 PM ANR0101E bfdedup.c(8813): Error 11 opening table "BF.Dereferenced.Chunks". 11/8/2010 9:55:34 PM ANR0101E bfdedup.c(9533): Error 11 opening table "BF.Queued.Chunks". 11/8/2010 9:55:36 PM ANR0101E bfdedup.c(9533): Error 11 opening table "BF.Queued.Chunks". 11/8/2010 9:55:37 PM ANR0101E bfdedup.c(9533): Error 11 opening table "BF.Queued.Chunks". 11/8/2010 9:55:39 PM ANR0101E bfdedup.c(9533): Error 11 opening table "BF.Queued.Chunks". I can only guess from the module name is it was trying to perform dedup processing. This is strange since none of the storage pools are defined as dedup. How can I turn off dedup processing? I also saw this message: 11/8/2010 10:16:15 PM ANR0130E dbieval.c(891): Server LOG space exhausted. which is also a little odd since I have 60GB specified for the active log and the archlog filesystem has >300GB free which was available when they called at 2am. It should be noted that I also turned back on ALLOWREORGTABLE (which I promptly turned off before restarting the server), thinking the previous problems with this option had been resolved and I had adequate archlog space. Any ideas?
