We experienced a repserver "crash" because repserver tried to allocate
memory, but reached the memory_limit. However this seems strange, because we
know that the replication server has been handling a very low amount of
transactions (maybe 2 per minute!) and also that these transactions are
normally very small.
The replication is a warm standby configuration replicating only one
database, running on the standby node of the two HP-UX 11.11 servers.
Replication server version: Replication Server/12.6/EBF 12238
ESD#3/HP9000-889/HP-UX 11.00/1/OPT/Mon Oct 4 04:39:12 2004
Dataserver version: Adaptive Server Enterprise/12.5.0.1/EBF 10428
ESD#2/P/HP9000-879/HP-UX 11.0/rel12501/1792/64-bit/FBO/Tue Jul 30 08:20:25
2002
This is last part of the log.
I. 2005/10/01 02:54:23. Connection to server 'SYB_SDP1' as user 'rep' has
been faded out (closed).
T. 2005/10/01 05:45:58. (119): Additional allocation of 65584 bytes to the
currently allocated memory of 209663893 bytes would exceed the memory_limit
of 209715200 specified in the configuration.
F. 2005/10/01 05:45:58. FATAL ERROR #7035 DSI EXEC(124(1)
SYB_SDP1.sdp_db1) - /dsiutil.c(2035)
Additional allocation would exceed the memory_limit of '209715200' specified
in the configuration.
T. 2005/10/01 05:45:58. (119): Exiting due to a fatal error
Now since there is almost no activity I watched the memory usage of
repserver and found that it is slowly increasing. Sometimes I see jumps in
memory usage, but it always goes up, never down. IS this intended behaviour
or is this maybe a memory leak??
Thanks in advance,
René van Deursen