Bug #15603 | Test failures 'im_daemon_life_cycle'. 'im_life_cycle, 'im_options_set' | ||
---|---|---|---|
Submitted: | 8 Dec 2005 21:10 | Modified: | 19 Dec 2005 13:45 |
Reporter: | Kristian Nielsen | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Server | Severity: | S3 (Non-critical) |
Version: | clone-5.0.17-build | OS: | IBM AIX (AIX, OpenBSD, HPUX11, Solaris 10) |
Assigned to: | Kristian Nielsen | CPU Architecture: | Any |
[8 Dec 2005 21:10]
Kristian Nielsen
[9 Dec 2005 8:19]
Valeriy Kravchuk
Thank you for a problem report. What exact ChangeSet was used? Or, in other words, what sources should I download to repeat the problem you encountered?
[12 Dec 2005 8:04]
Kristian Nielsen
This bug seems to be a defect in the test suite/build tools rather than a 'real' bug in the server code. Two suggestions: 1. Print out all used ports and sockets, check for conflicts. 2. Increase sleeps here, see if it helps, indicating a race condition: im_life_cycle.imtest:38:--sleep 3 im_life_cycle.imtest:66:--sleep 3 im_life_cycle.imtest:136:--sleep 3 im_utils.imtest:45:-- sleep 3 Note: Keep petr informed about progress on this bug.
[13 Dec 2005 8:41]
Kristian Nielsen
I looked closer at the test log. This is how the two parallel tests are run on host aix52: perl ./mysql-test-run.pl --tmpdir=/home/mysqldev/tmp/my_build-aix52-64bit --master_port=3336 --slave_port=3352 --manager-port=3338 --no-manager --timer --sleep=10 --force --im-port=3339 --im-mysqld1-port=3341 --im-mysqld2-port=3343 perl ./mysql-test-run.pl --tmpdir=/home/mysqldev/tmp/my_build-aix52 --master_port=3334 --slave_port=3350 --manager-port=3336 --no-manager --timer --sleep=10 --force --im-port=3337 --im-mysqld1-port=3339 --im-mysqld2-port=3341 So the manager port for the second run is the same as the master port for the first run. This clearly cannot work. We need to fix the build scripts to not use the same ports for simultaneous test runs!