G
George Potemkin
Guest
Carl, do you always pre-start 60 remote client servers immidiately after db startup? [2016/07/13@23:48:13.123+0200] P-3986 T-140605298083648 I APW 82: (2518) Started. [2016/07/13@23:48:27.788+0200] P-4350 T-140402236524352 I SRV 1: (5646) Started on port 11002 using TCP IPV4 address 0.0.0.0, pid 4350. [2016/07/13@23:49:14.562+0200] P-4843 T-140184589739840 I SRV 2: (5646) Started on port 11003 using TCP IPV4 address 0.0.0.0, pid 4843. [2016/07/14@00:26:16.996+0200] P-17010 T-140595944453952 I ABL 87: (1132) Invalid semaphore id [2016/07/14@00:26:16.996+0200] P-17010 T-140595944453952 I ABL 87: (10839) SYSTEM ERROR: Unable to set semaphore set : semid = 7, errno = 22. [2016/07/14@00:26:16.996+0200] P-17010 T-140595944453952 I ABL 87: (2252) Begin transaction backout. [2016/07/14@00:26:16.996+0200] P-17010 T-140595944453952 I ABL 87: (2253) Transaction backout completed. [2016/07/14@00:26:17.002+0200] P-17010 T-140595944453952 I ABL (1132) Invalid semaphore id [2016/07/14@00:26:17.002+0200] P-17010 T-140595944453952 I ABL (1132) Invalid semaphore id [2016/07/14@00:26:17.002+0200] P-17010 T-140595944453952 I ABL (10839) SYSTEM ERROR: Unable to get value of semaphore set : semid = 5, errno = 22. [2016/07/14@00:26:17.002+0200] P-17010 T-140595944453952 I ABL (-----) semLockLog_2: semValue val = -1 [2016/07/14@00:26:19.903+0200] P-3966 T-140549975136064 I BROKER 0: (2527) Disconnecting dead user 87. [2016/07/14@00:55:14.379+0200] P-2288 T-139877706889024 I APW 82: (2518) Started. [2016/07/14@00:57:04.277+0200] P-3010 T-140641843181376 I SRV 2: (5646) Started on port 11003 using TCP IPV4 address 0.0.0.0, pid 3010. ... [2016/07/14@01:02:05.506+0200] P-4201 T-140565902128960 I SRV 5: (5646) Started on port 11006 using TCP IPV4 address 0.0.0.0, pid 4201. [2016/07/14@01:25:33.654+0200] P-2626 T-139742344922944 I APPSRV 91: (1132) Invalid semaphore id [2016/07/14@01:25:33.654+0200] P-2626 T-139742344922944 I APPSRV 91: (10839) SYSTEM ERROR: Unable to set semaphore set : semid = -32761, errno = 22. It looks like sometimes only a few servers were started successfully and the semaphore errors were issued by the self-service clients: Latch activity collected by dbmon is normal. And I guess db activity was next to zero right after db startup in the midnight. I suspect that starting many RCS at once can be a root case of the errors.Multiple semsets will decrease the competion for semset's during their launches.
Continue reading...
Continue reading...