Comment 10 for bug 1250643

Revision history for this message
vincent.seguin (vincent-seguin) wrote :

@Shahriyar I don't think you'll be able to reproduce easily without the actual database

Out of >800 recent restores for this particular version combination (different databases, with similar my.cnf), I have 3 confirmed "bad" backups that can't be restored with that problem.

The only thing I know is that:
1. I have seen it on mysql 5.5.36 only
2. It doesn't appear to occur on 5.6 (>1600 restores with no issue)
3. I have seen it with xtrabackup 2.2.10 and 2.2.12
4. It only occurs on differential backups
5. The 3 cases I've seen are on large databases (>2T database size)
6. In the most recent case, two successive differentials of the same DB (same base LSN) exhibit the same problem
7. A new full backup, then differential on the same DB can be restored just fine (confirming in the next couple days)

The command that fails is:

innobackupex --apply-log --incremental-dir /glide/mysqld/aztemptesting_3400_giga/temp/restore_2016-02-01_1336089/azprod_3403_db160098_37-2_2016-
01-31_1844124 --use-memory=32G /glide/mysqld/aztemptesting_3400_giga/temp/restore_2016-02-01_1336089/azprod_3403_db160098_37_2016-01-29_1844126 2> /glide/backup/log/restore.2
016-02-01_1336089.azprod_3403_db160098_37-2_2016-01-31_1844124.log