MySQL Bugs: #32998: epoch on slave is much higher (20 digits) than epoch on master (8 digits)

Bug #32998	epoch on slave is much higher (20 digits) than epoch on master (8 digits)
Submitted:	5 Dec 2007 13:36	Modified:	11 Dec 2008 14:16
Reporter:	Oli Sennhauser	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Cluster: Replication	Severity:	S3 (Non-critical)
Version:	5.1.22	OS:	HP/UX (HP-UX B.11.11 U 9000/800)
Assigned to:		CPU Architecture:	Any

Description:
We are doing a Cluster -> InnoDB replication on HP-UX. When we want to do a channel failover from the slave to the other master the last applied epoch on the slave is a 20 digit number on the slave and the according query on the master is a 8 digit number for the ndb_binlog_index.

How to repeat:
set up replication and test that it is running.

do a select * from ndb_apply_status on the slave
do a select * from ndb_binlog_index on the master

compare the epoch fields

Suggested fix:
maybe a endian problem or a 32/64-bit problem?

Hi Oli, 

i tested this on our hpux11 machine (HP/UX 11.11 pa-risc 2.0 64bit) using 64bit 5.1.22 binaries and could not reproduce it there.

Did you use a mixed 32/64 bit setup in your tests?

Can't reproduce with replication from 64bit mysqld -> 32bit mysqld either ...

I'm guessing this is CGE and not 5.1.22?

if do epoch numbers can are big because of the microGCP's...

please let us know all the versions of the respective clusters involved.

BR,

Tomas

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

Oli,

we still needs the feedback from you.

Susanne

I am not at the customer anymore so I cannot provide any information. But I pretty sure it was NOT CGE!