stable

Clone or download

Read-only

Tuleap init.d script should start backend workers as well

TL;DR system events cannot reliably start workers 'cause: - either the process dies and because Zombie (in your head) - or the process doesn't die and keep the system_event running (in your a**) The bug was always there but was hidden by the fact that backend workers reconnection with redis was broken so the workers always failed but were always restarted next time (cron processing). As request #13893 fixed that with commit 8ef9da3cdb45742126e4872f36c0ce1f021052cd we can now totally savor the "don't rewrite your own init system" cake. Change-Id: I28d47d0cbd1d43f932a107bba18497e86ae41de1

Modified Files

Name
A src/common/CLI/Command/WorkerSupervisorCommand.php +252 −0 Go to diff View file
A src/common/CLI/Command/WorkerSystemCtlCommand.php +116 −0 Go to diff View file
M src/common/Queue/Worker.php +17 −22 Go to diff View file
M src/common/system_event/SystemEventProcessor_ApplicationOwner.class.php +2 −23 Go to diff View file
M src/utils/init.d/codendi +13 −22 Go to diff View file
A src/utils/systemd/tuleap-worker@.service +9 −0 Go to diff View file
A src/utils/systemd/tuleap-workers.service +13 −0 Go to diff View file
M src/utils/systemd/tuleap.service +1 −0 Go to diff View file
M src/utils/tuleap.php +22 −0 Go to diff View file
M tests/build_and_run/test.sh +4 −1 Go to diff View file
M tools/rpm/build_and_run_packages.sh +1 −1 Go to diff View file
M tools/rpm/tuleap.rhel7.spec +6 −0 Go to diff View file