Termination with the following syslog messages: DBI LOG (BZ3): invalid return code OLF received.
DBI LOG (BZE) Invalid SQL-Error 0
DBI LOG (BY2) SQL-Error 0 occured while performing COM
OLF == optimistic locking failed
You activated the optimistic writting of cluster records, that is the cluster interface works with a so-called optimistic approach, it generally does not generate locks and delays writing operations to the last possible date. This assumes that no other process modifies the records that are going to be changed in the meantime.
When writing cluster records, a time stamp is always maintained as well (field TIMESTMP in the physical cluster table). The check to see whether the affected records have been changed uses this time stamp.
The cluster interface remembers the time stamp of the cluster (field TIMESTMP) when it is imported and supplements the update/delete operations correspondingly ( .. AND TIMESTMP = old time keys). If the update/delete operation finds nothing, this means that another process has changed the cluster in parallel.
Another process has obviously changed the cluster records in parallel.
By deactivating the optimistic writting you can avoid the terminations. Instead, you must set the profile parameter rsdb/rclu/optimistic = 0 in Releases up to and including 3.0C.
In Releases larger than 3.0C this parameter was replaced by a new parameter rsdb/rclu/opt_level. The following settings are possible to deactivate the optimistic writting here:
rsdb/rclu/opt_level = 0 (optimistic reading is inactive) or
rsdb/rclu/opt_level = 2 (optimistic reading is active).
No comments:
Post a Comment