pt-table-sync + stock mysql 5.5.25a error

[Brian Fraser]

This is for the second bug reported in https://bugs.launchpad.net/percona-toolkit/+bug/1034717

[Brian Fraser (fraserbn)]

While I'm still unable to reproduce this, I just pushed a branch with a possible fix; it removes some special-casing for the latin1 range that might've been causing this. Ryan, since right now you are my test case, would you mind checking if that works for you?

[Ryan Brothers (ryan-brothers)]

Brian - thanks for your help. I confirmed that your fix resolves the issue for me in MySQL 5.5.25a.

[Brian Fraser (fraserbn)]

Fantastic! However, the more I dive into the chunking code, the more misgivings I have, so I'm not going to merge this in for the 2.1.4 release; instead, I'll tag it for 2.1.5, and take a bit more of time to poke things.

[Brian Fraser (fraserbn)]

Talked about this with Daniel, and the consensus was that it's not worth merging into the 2.1 series as it might break something, and the problem isn't too widespread. It *might* make it into the 2.2 series, but more than likely it'll simply stop being a problem once we merge pt-table-checksum and pt-table-sync. Until then, people with this issue will have to make do with using this version of pt-table-sync.

[Ryan Brothers (ryan-brothers)]

Thanks for the update. I'm fine with this being delayed (or not being released at all). I ended up just switching to a different algorithm to resolve the issue - using Nibble instead of Chunk.

[Daniel Nichter (daniel-nichter)]

I'm going to close this as WontFix not because we won't fix it but because we're going to rewrite pt-table-sync entirely, so I'm not sure how this will fit. In any case, we have the branch and this issue will be archived so we can revisit once pts is rewritten.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PT-1016