Re: Struggling to get replication up...

[Bob Bankay]

Hi Yasir,

The corruption is because the query had a special character in it and it is either a records delimiter or something in your query that go logged incorrectly:

Query: 'insert into productdescription (createdate,
createdbyid, onelinedescid, productid, languageId,marketid,description,
isactive, updatedate, dbuäDAB', Error_code: 1064
----------------------------------^^------------------------------- that is confusing the sql reader on the slave.

We had a similar problem when we ran out of space for binlog for about 15 minutes. The binlog got corrupted.
We were unable to find a valid start point within the corrupt binlog. So we pointed the slave at the following binlog.nnn+1 offset 4 and restarted the slave.

When we be to a scheduled outage we will remigrate the data to the slave.

Finally since the binlog interpreter read the corrupt data it also give the errorr that the packet size was too small. Since it is written in binary code, it could no longer find the end record markers.

Don't know it this helps.