Test automation using juju with LE (1.24.5) is looking solid. I have yet to observe a single failure-to-cluster occurrence. I've cycled 50+ on bare metal and NaN virtually. ;-)
Manual testing using juju sans LE (1.22.6) is significantly improved.
I did some non-LE iterations on bare metal last week which showed only a rare failure-to-cluster occurrence (2 out of 53 loops). That, versus a complete failure (25 of 25) to cluster without the proposed cluster fixes when using pre-LE jujus + rmq 15.07 charm. FWIW, that was on bare metal, with all hosts resolving forward and reverse.
+$0.02:
I've got this branch, plus the native- cluster- race-fixes branch both merged into my MP for the refactored and extended rmq amulet tests @
https:/ /code.launchpad .net/~1chb1n/ charms/ trusty/ rabbitmq- server/ amulet- refactor- 1508/+merge/ 269749
With that:
Test automation using juju with LE (1.24.5) is looking solid. I have yet to observe a single failure-to-cluster occurrence. I've cycled 50+ on bare metal and NaN virtually. ;-)
Manual testing using juju sans LE (1.22.6) is significantly improved.
I did some non-LE iterations on bare metal last week which showed only a rare failure-to-cluster occurrence (2 out of 53 loops). That, versus a complete failure (25 of 25) to cluster without the proposed cluster fixes when using pre-LE jujus + rmq 15.07 charm. FWIW, that was on bare metal, with all hosts resolving forward and reverse.