Closed dcbw closed 8 years ago
@danwinship good point; updated. HA stuff must use a config file at this point, since the master controller service needs it. But non-HA definitely can just use command-line args.
@danwinship @pravisankar PTAL; now each config file (if it exists) is added to the tarball along with its service. The 'zzzz' seems necessary since when passing an array through to a bash function the array gets expanded and empty elements won't be present, so otherwise the indexes would no longer be the same between service_names and config_files. But that may also just be my bash-inexperience. Still, it produces the right result for me...
@danwinship @pravisankar @knobunc PTAL, thanks.
Yes, the branding/services may change, but we'll have to update the script for that anyway...
assuming it's tested, LGTM
@danwinship I ran it on an HA cluster and it spits out the right logs. I haven't run it on a non-HA cluster though, but I'll do that today.
FWIW, the log output now looks like this...
./master/journal-atomic-openshift-master.service ./master/systemctl-show-atomic-openshift-master.service ./master/journal-atomic-openshift-master-controllers.service ./master/systemctl-show-atomic-openshift-master-controllers.service ./master/CONFIG-atomic-openshift-master-controllers.service ./master/journal-atomic-openshift-master-api.service ./master/systemctl-show-atomic-openshift-master-api.service ./master/CONFIG-atomic-openshift-master-api.service
and for a non-HA master:
openshift-sdn-debug/master/journal-openshift-master.service openshift-sdn-debug/master/systemctl-show-openshift-master.service
See discussion on #287 by @knobunc. This has the same problem, that it assumes there will be a config file for the master, but some deployments (including, I believe, our default non-ha ansible deployment) don't have any such file; they just use command-line arguments.