Using git-buildpackage for continuous integration we have
something like:
| /usr/bin/make DESTDIR=/tmp/buildd/ngcp-asterisk-1.4.24.1+ngcp.1+0~1367942255.82+squeeze~1.gbp58b891/debian/ngcp-asterisk/ install
in some of our build processes. To avoid failing at this stage
modify the check in the "badshell" make target to check for "~/"
instead of just checking for "~" anywhere inside the version
string.
Also see: https://issues.asterisk.org/jira/browse/ASTERISK-13797
Closes: https://jenkins.mgm.sipwise.com/job/asterisk-binaries/78/
Otherwise it looks like a package with orig sources sitting in
a foreign tree to git-buildpackage:
| + git-buildpackage -nc --git-force-create --git-ignore-new -S -us -uc --git-verbose --git-builder=/bin/true --git-cleaner=/bin/true
| /bin/true [] []
| fatal: Not a valid object name upstream/1.4.24.1
| ngcp-asterisk_1.4.24.1.orig.tar.gz does not exist, creating from 'upstream'
Moving to this version number schema instead fixes the
build issue.
Closes: https://jenkins.mgm.sipwise.com/job/asterisk-source/81/
This makes reviews and diffs easier and helps in avoiding duplicate entries.
devscripts's wrap-and-sort ftw.
Closes: https://bugtracker.sipwise.com/view.php?id=1845
From: Michael Prokop <mprokop@sipwise.com>
Somehow looks like asterisk does not like a start
when it'sa already started. I never had any
issues but read about it. It's safe to check
status before running start.
Still need to work in asterisk restart upon
upgrades. And how to deal with monit for that.
Until now, the notification mail was using user's
UUID in the message. Now it will use main E164
number.
Another VM_UUID var is exported if UUID is required.