I recently changed connect_limit (doubled it) and after that change
I got transaction log every morning full. It happened during the night
I suspected  that it may be one of the cronjobs running, so I
disabled
for the weekend all nightly crons that update the databases.

I haven't change anything else. I saw some of the parameters were
automatically
Reconfigered after changing connect_limit - not all parameters were
doubled, increased but not doubled.

I didn't think this parameter should affect the transaction log file.
Am I right?

I could increase the size of the transaction log file but it is good
to know the reason
for filling it. It is recommended (in errlog.log) to reduce logfull
and force_abort
limits to give the Logging System more free logspace  for recovery.
Currently logfull is 90%, force_abort - 72%. What are the best numbers
for those
2 parameters.

I think archiver works fine (I read this my be a problem simetimes).

The other think - I just came from maternity finding that Replicator
was installed
(Ingres 2.6/0305) between two db servers. Could this  cause the
problem?
45 minutes before LOGFULL error I see error message related to the
replicator:

yoda    ::[45051        IIGCC, 00000001]: Mon Feb 19 05:24:29 2007
E_GC2205_RMT_ABORT   Session failure: ABORT received from remote
partner.
yoda    ::[45051        IIGCC, 00000001]: Mon Feb 19 05:24:29 2007
E_CLFE07_BS_READ_ERR Read from peer process failed; it may have
exited.
YODA    ::[45049             , 0000000103ea0200]: Mon Feb 19 05:35:17
2007 W_DM956A_REP_MANUAL_DIST   Outstanding record(s) were found in
the replication input queue when opening database , this was probably
due to a system crash. These record(s) will be processed manually by
the current user thread.
yoda    ::[45051        IIGCC, 00000002]: Mon Feb 19 05:46:04 2007
E_GC2205_RMT_ABORT   Session failure: ABORT received from remote
partner.
yoda    ::[45051        IIGCC, 00000002]: Mon Feb 19 05:46:04 2007
E_CLFE07_BS_READ_ERR Read from peer process failed; it may have
exited.
YODA    ::[45049             , 0000000103ea0200]: Mon Feb 19 06:17:33
2007 E_DMA48E_LOGFULL_EXCEEDED
  Unexpected condition during System Logfull Handling - the Logfile
has grown in size beyond the logfull point.  The combined used and
reserved space in  the transaction Log File has grown to exceed what
the Logging System has  saved for its recovery handling.  This may
indicate that the Logging System  has not reserved enough logfile
space for rollback processing, or it may  indicate that concurrent
activity in the system has unexpectedly used up  logfile space more
quickly than expected.  The Logging System will be  shut down so that
recovery can be run in offline mode where logfile space  usage is more
controlled.  If this condition recurs then reduce the system  logfull
and force-abort limits to give the Logging System more free logspace
for recovery.  Log File BOF (1156250381, 41420, 292).  Log File EOF
(1156250381, 72600, 1816).   Reserved space: 180564044.

Any hints?

Thanks,
Nina Markova