Merge lp:~jtv/maas/bug-1086239 into lp:maas/trunk
| Status: | Merged |
|---|---|
| Approved by: | Julian Edwards on 2012-12-12 |
| Approved revision: | 1399 |
| Merged at revision: | 1397 |
| Proposed branch: | lp:~jtv/maas/bug-1086239 |
| Merge into: | lp:maas/trunk |
| Diff against target: |
393 lines (+111/-68) 14 files modified
etc/democeleryconfig_cluster.py (+0/-4) services/cluster-worker/run (+5/-0) services/pserv/run (+5/-0) src/provisioningserver/auth.py (+0/-8) src/provisioningserver/boot_images.py (+5/-5) src/provisioningserver/cluster_config.py (+44/-0) src/provisioningserver/dhcp/leases.py (+2/-2) src/provisioningserver/dhcp/tests/test_leases.py (+0/-9) src/provisioningserver/start_cluster_controller.py (+1/-8) src/provisioningserver/tags.py (+2/-6) src/provisioningserver/tests/test_auth.py (+0/-12) src/provisioningserver/tests/test_boot_images.py (+0/-10) src/provisioningserver/tests/test_cluster_config.py (+46/-0) src/provisioningserver/tftp.py (+1/-4) |
| To merge this branch: | bzr merge lp:~jtv/maas/bug-1086239 |
| Related bugs: |
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Gavin Panella (community) | 2012-12-11 | Approve on 2012-12-11 | |
|
Review via email:
|
|||
Commit Message
Take cluster-uuid setting from environment, not celery config.
Description of the Change
This builds on a multi-step change agreed with Julian. The relevant part is:
1 Duplicate the CLUSTER_UUID setting, from Celery config to maas_cluster.conf.
2 Make pserv's upstart job source maas_cluster.conf.
3 Have get_cluster_uuid read the setting from the environment instead of from celery.
4 Clean up the duplicated setting.
What you see here is step 3. It also puts get_cluster_uuid in a saner place. That function was "borrowed" from a place where nobody should be importing it from. Not so bad for a provisional fix, but this is the cleanup.
Finally, the cluster uuid is no longer one of the "worker secrets" that the region controller sends to the clusters. So this also removes some of the remnants of that old situation. In particular, "I don't know my cluster uuid yet" is no longer a valid reason for cluster controllers to be unable to perform tasks. Now, if the uuid is not set, either installation failed in some unusual way or we're running something that needs the information without sourcing maas_cluster.conf. There may be some nooks and crannies where this may still need doing, so I tried to make the failure a helpful one.
All of this work is to be backported to the 1.2 branch.
Jeroen
| MAAS Lander (maas-lander) wrote : | # |
| Jeroen T. Vermeulen (jtv) wrote : | # |
Weird Jenkins failure. I'll try again. I also snuck in a small change for the part I forgot to cover: running a demo cluster controller in a source branch.
| MAAS Lander (maas-lander) wrote : | # |
The Jenkins job https:/
Not merging it.


The Jenkins job https:/ /jenkins. qa.ubuntu. com/job/ maas-merger- trunk/129/ console reported an error when processing this lp:~jtv/maas/bug-1086239 branch.
Not merging it.