Raise an error when transaction log wraps around before being processed by xtrabackup
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Percona XtraBackup moved to https://jira.percona.com/projects/PXB |
Fix Released
|
High
|
Alexey Kopytov | ||
2.0 |
Fix Released
|
High
|
Alexey Kopytov |
Bug Description
Currently when xtrabackup discovers that the transaction log has wrapped around before all log records have been copied (that is, the log copying thread is too late due to either load or the log files being too small) , xtrabackup prints the following message and the continues without failures (just printing the same message over and over until the log copying thread is terminated):
"
>> ###Warning###: The copying transaction log migh be overtaken already by the target.
>> : Waiting log block no 16732, but the bumber is already 20820.
>> : If the number equals 16732 + n * 4088, it should be overtaken already.
"
What it should do instead is to make the entire backup process fail, because hitting that condition can never result in a valid backup.
Related branches
- Stewart Smith (community): Approve
-
Diff: 168 lines (+86/-42)3 files modifiedtest/run.sh (+1/-1)
test/t/xb_log_overwrap.sh (+38/-0)
xtrabackup.c (+47/-41)
affects: | percona-server → percona-xtrabackup |
tags: | added: cr i17015 |
I'd like to add that a more verbose error message how to increase the innodb_ log_file_ size would be nice.