HP 3000 Manuals HP 3000 Manuals
Error Messages during Soft Resynchronization
ALLBASE/Replicate produces either warnings or error messages as you start
soft resynchronization. These indicate the status of each partition
being replicated. Review these error messages and warnings to determine
whether hard resynchronization is needed. Some sample messages and their
meaning are shown below:[REV END]
* openlog: Partition 15 has no committed transactions
* This message is just a warning, telling you that no
transactions have yet committed against master partition 15
for replication on the slave. Soft Resynchronization will
continue for partition 15, and it will send transactions to
the slave as soon as any commit on the master.
* openlog: Hard Resynch required for partition 10
* This message indicates that a hard resynchronization is
necessary for partition 10. This condition could arise
because you did a soft resynchronization from a slave that
had been previously hard resynched from another master.
* openlog: Insufficient log history for partition 23.
* This message indicates that partition 23 has been found,
but the master log records do not go far enough back to
contain the last transaction sent to the slave. This
condition probably exists because the log files filled and
were allowed to overwrite the previous log records. This
condition requires a hard resynchronization.
* openlog: Slave is ahead of the master for partition 3.
* This message indicates that the transaction identifier for
the latest transaction committed to the slave is higher
than the transaction identifier for the last transaction
committed on the master for partition 3. This condition
occurred because direct updates have been done to tables on
the slave that are contained in partition 3 on the master.
This usually occurs when the slave and the master have
switched roles, and updates have been done to tables on the
slave in its role of new master. The old master was put
back in service as master without applying the updates made
to the slave, leaving the slave ahead of the master.
This problem can occur any time direct updates are made to
the slave, and the slave has the same partition number as
the master. If the resynchronization application is
started after the direct update is made, the application
will fail due to this condition.
This condition can be remedied by doing a soft
resynchronization from the slave (new master) back to the
old master -- if there is not a huge backlog of committed
transactions on the slave (new master). If the backlog is
large enough, a hard resynchronization may be desirable.