Merge lp:~akopytov/percona-xtrabackup/bug1097158-2.1 into lp:percona-xtrabackup/2.1
Status: | Merged | ||||
---|---|---|---|---|---|
Approved by: | Stewart Smith | ||||
Approved revision: | no longer in the source branch. | ||||
Merged at revision: | 475 | ||||
Proposed branch: | lp:~akopytov/percona-xtrabackup/bug1097158-2.1 | ||||
Merge into: | lp:percona-xtrabackup/2.1 | ||||
Diff against target: |
76 lines (+23/-13) 2 files modified
src/xtrabackup.c (+3/-0) test/t/xb_log_overwrap.sh (+20/-13) |
||||
To merge this branch: | bzr merge lp:~akopytov/percona-xtrabackup/bug1097158-2.1 | ||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Stewart Smith (community) | Approve | ||
Registry Administrators | Pending | ||
Review via email: mp+142292@code.launchpad.net |
Description of the change
Bug #1097158: xb_log_overwrap fails sporadically in Jenkin
The test case tries to generate enough log writes to cause the log files
wrap around and make xtrabackup fail with an error.
Found 2 potential problems:
- the test used the default log buffer size. Which, for an InnoDB
plugin's default 8M buffer size, means there may be not enough changes
in the on-disk files to trigger the expected error
- the test suspended the main xtrabackup thread, but not the log copying
thread. Which means the log copying thread could be catching up in the
background and avoid reading wrapped-around log entries, and thus, the
expected error.
Fixed by setting an explicit log buffer size to the minimum value (1M)
and using debug sync to suspend log copying thread, rather than the main
one.
http://