Make MTR catch server error log mishandling automatically
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Percona Server moved to https://jira.percona.com/projects/PS | Status tracked in 5.7 | |||||
5.5 |
Triaged
|
Wishlist
|
Unassigned | |||
5.6 |
Triaged
|
Wishlist
|
Unassigned | |||
5.7 |
Triaged
|
Wishlist
|
Unassigned |
Bug Description
- For each worker, note the 1st test name running there. After each testcase, head -n 1 the server error log for "CURRENT_TEST: foo" in order to catch the testcases that delete the error log
- For mysql-test/include/ grep helpers, die if the file being searched is the error log in its default name and not e.g. $assert_only_after is set in case of assert_grep.inc. This would catch false negatives by finding output of previous testcases on the same worker. Check if $assert_only_after and friends actually work with --repeat.
- In testcase warning checker, list tmp and log dirs before and after a testcase run and add the diff to the warning checker output, to catch stray custom server error logs (which again break --repeat and possibly other testcases if the name happens to be shared)
tags: | added: ci upstream |
The 1st item could be hacked in an unportable way by marking error logs as append-only (chflags unappnd on mac, chattr +a on Linux)