Bug 107052
| Summary: | Buildmaster shouldn't restart after all webkitpy change | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Csaba Osztrogonác <ossy> |
| Component: | New Bugs | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED WONTFIX | ||
| Severity: | Normal | CC: | lforschler, ossy, wsiegrist |
| Priority: | P2 | ||
| Version: | 420+ | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
Csaba Osztrogonác
It seems build.webkit.org restarts itself after all webkitpy change.
Today it was restarted twice in a short period because of these changes:
- http://build.webkit.org/changes/20528 (http://trac.webkit.org/changeset/139913)
- http://build.webkit.org/changes/20539 (http://trac.webkit.org/changeset/139924)
Do we really need restarting master after all webkitpy change?
I think only master.cfg changes should trigger master restart.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Lucas Forschler
Buildbot is restarted via post commit hook from the svn sever. The post commit hook reads as follows:
#
# update build.webkit.org
#
UPDATE_BUILDBOT=$(/usr/bin/svnlook changed $REPOS -r $REV | /bin/grep -cE '(webkitpy|BuildSlaveSupport)')
if [ $UPDATE_BUILDBOT -gt 0 ];
then
/usr/bin/ssh -t buildbot@build.webkit.org "/usr/bin/sudo -n /usr/local/bin/buildbot-update-restart --force" >> /tmp/post-commit.log 2>&1 &
fi
Lucas Forschler
However, looking in the buildbot-update-restart code, it seems we filter out on config.json or master.cfg changes
RESTART=$(/usr/bin/svn log -r $OLDREV:HEAD -vq | grep -cE '(config.json|master.cfg)');