~mfoord/juju-core/write-majority : changes

~mfoord/juju-core/write-majority » Changes from revision 2747

From Revision 2747 to 2728

Rev	Summary	Authors	Date
2747	Initial merge Initial merge	Michael Foord	9 years ago
2746	Merge Merge	Michael Foord	9 years ago
2745	[r=klyachin],[bug=1309444] HA: peergrouper spins in local en... [r=klyachin],[bug=1309444] HA: peergrouper spins in local environment The peergrouper is intended to monitor the MongoDB replica health and make sure everything is functioning well. However, when run in the local provider, or in an upgraded environment (while we haven't finished all the upgrade steps), it spams the log file with: INFO juju.worker runner.go:260 start "peergrouper" ERROR juju.worker.peergrouper worker.go:137 peergrouper loop terminated: cannot get replica set status: cannot get replica set status: not running with --replSet This fix ensure NOT running of peergrouper worker in local environment. https://codereview.appspot.com/97880045/ R=axwalk, fwereade, jameinel, klyachin R=axwalk, fwereade, jameinel, klyachin	Vladislav Klyachin	9 years ago
2744	[r=hduran-8] Added explicit --journal flag to mongod [r=hduran-8] Added explicit --journal flag to mongod Mongod has Write Ahead logging on by default for 64 bits systems: http://docs.mongodb.org/manual/tutorial/manage-journaling/ juju seems to be assuming this for every arch so the specific --journal flag was added. https://codereview.appspot.com/97370044/ R=fwereade, hduran-8, wwitzel3	Horacio Durán	9 years ago
2743	[r=fwereade],[bug=1320138] Fix LP 1320138 [r=fwereade],[bug=1320138] Fix LP 1320138 Constraints appear to be a set, not a list, so compare them in an unordered manner. https://codereview.appspot.com/94520045/	Dave Cheney	9 years ago
2742	[r=axwalk],[bug=1262967] Fix tests on i386 [r=axwalk],[bug=1262967] Fix tests on i386 Various fixes to run tests on i386. For Joyent, since it only supports AMD64 I have modified the suites to upload fake AMD64 tools; this involved a change to envtesting.ToolsFixture to parameterise UploadFakeTools. For OpenStack, I added dummy metadata for i386 and ppc64. Fixes lp:1262967 https://codereview.appspot.com/92390046/	Andrew Wilkins	9 years ago
2741	[r=axwalk] Fix arch preference algorithm [r=axwalk] Fix arch preference algorithm Change the instance architecture preference selection to widest word-size, then first arch name by alphabetical order. For tie- breakers, choose the instance type based on provider-specific ordering. In the future we may want to extend the ordering to include cost. For now, we rely on the providers ordering instance types by their own preference (no change here). https://codereview.appspot.com/94530045/	Andrew Wilkins	9 years ago
2740	[r=thumper] update errgo [r=thumper] update errgo Fixes gccgo incompatability	Dave Cheney	9 years ago
2739	[r=jameinel] utils/registry: TypedNameVersion registry [r=jameinel] utils/registry: TypedNameVersion registry As part of getting API versioning in place, I was putting together some Registry's in a couple of places. I realized I wanted 2 things that were essentially identical code, just tracking slightly different object types. So I put together an abstraction that still enforces the Type, but is otherwise treats the input and output as a generic interface{}. https://codereview.appspot.com/99180044/	John Arbash Meinel	9 years ago
2738	[r=dave-cheney] provider/openstack: upgrade goose [r=dave-cheney] provider/openstack: upgrade goose	Dave Cheney	9 years ago
2737	[r=axwalk] replicaset: close cloned sessions [r=axwalk] replicaset: close cloned sessions Not sure if this is related to test failures, but we should close cloned sessions anyway. https://codereview.appspot.com/92400043/	Andrew Wilkins	9 years ago
2736	[r=wallyworld] New test suite: BaseSuite [r=wallyworld] New test suite: BaseSuite A new base test suite is introduced. It is to replace LoggingSuite as the preferred suite which all test suites should encapsulate. BaseSuite itseld encapsulates LoggingSuite, plus provides: - isolation of user home - flag to ensure tests will error if outgoing network access is attempted Much of the implementation was moved across from testing/environs.go. There were a number of slightly different methods to create a fake Juju home setup - these have been consolidated and simplified. The flag to turn off outgoing network access when running tests is off until the tests are fixed. Not all tests have been updated to use the new suite; only those affected by the refactoring have been fixed for now. https://codereview.appspot.com/100460044/	Ian Booth	9 years ago
2735	[r=axwalk],[bug=1319617] replicaset: retry replSetInitiate [r=axwalk],[bug=1319617] replicaset: retry replSetInitiate It would seem there's a race in mongo, where it may think that the one and only local replicaset member is not reachable shortly after starting up. I was able to reliably reproduce this error on the bot machine by running TestInitiateReplica in worker/peergrouper in a loop. With the retry loop, no problems. Fixes lp:1319617 https://codereview.appspot.com/97520046/	Andrew Wilkins	9 years ago
2734	[r=wallyworld],[bug=1319822] Fix lxc tests on trusty [r=wallyworld],[bug=1319822] Fix lxc tests on trusty Now that we use lxc cloning by default on trusty (unless explicitly disabled), some tests needed updating to accommodate that. https://codereview.appspot.com/99330043/	Ian Booth	9 years ago
2733	[r=fwereade] state/apiserver/upgrader: we don't need datadir [r=fwereade] state/apiserver/upgrader: we don't need datadir UnitUpgrader was a unique facade in that it was taking a "datadir" argument (only other one that does that currently is Client). However, it doesn't actually do anything with it, so lets not give it useless data that we have to manufacture. https://codereview.appspot.com/98250044/	John Arbash Meinel	9 years ago
2732	[r=wallyworld],[bug=1318485] Use lxc clone by default [r=wallyworld],[bug=1318485] Use lxc clone by default All providers now support configuring lxc clone support, not just local provider. The config attributes which previously were just for local provider now are available on all providers: lxc-clone lxc-clone-aufs lxc-clone, if unspecified, will be set to true if the underlying platform supports it. The test is for Ubuntu trusty or greater. The final check is done when the container manager is created. 1.19.2 used a setting lxc-use-clone. This is now removed in place of the universal lxc-clone setting. https://codereview.appspot.com/94410043/	Ian Booth	9 years ago
2731	[r=thumper] Extract errors package to github.com/juju/errors [r=thumper] Extract errors package to github.com/juju/errors Also introduces a dependency on github.com/juju/errgo and removes the github.com/errgo/errgo dependency. Adds additional functions for simple tracing and annotations of errors. https://codereview.appspot.com/93410043/	Tim Penhey	9 years ago
2730	[r=axwalk],[bug=1318527] testing: drop databases before each... [r=axwalk],[bug=1318527] testing: drop databases before each Mgo test Previously the store package had its own mgo suite, which was largely the same as the one in juju-core/testing. It also had code to drop the non-admin databases before each test case, which I did not carry over when migrating. This caused flakeyness. Fixes lp:1318527 https://codereview.appspot.com/94460045/	Andrew Wilkins	9 years ago
2729	[r=axwalk],[bug=1301467] testing: reattempt MgoInstance.run ... [r=axwalk],[bug=1301467] testing: reattempt MgoInstance.run if addr in use Mongod does not support listening on port 0 to acquire an ephemeral port, so our tests must choose a port which may then be allocated to another process. This CL adds code to reattempt the starting of mongod if the port is in use, up to a maximum of 5 attempts. https://codereview.appspot.com/94410046/	Andrew Wilkins	9 years ago
2728	[r=axwalk],[bug=1260171],[bug=1316272] Change StopInstances ... [r=axwalk],[bug=1260171],[bug=1316272] Change StopInstances to take []instance.Id StopInstances previously took []instance.Instance, which is unnecessary for most providers, and imposes an unnecessary cost for callers that only have IDs initially. We change StopInstances to The openstack provider's StopInstances is now slightly more expensive than before, requiring an additional call to Instances. This cost will disappear when we fix the security groups/open-port implementation. Drive-by: the maas provider now uses a bulk "release" in StopInstances. Live tested on ec2, local, virtual MAAS, canonistack, azure. I lack a Joyent account, but I think it should be fine. Fixes lp:1260171 Fixes lp:1316272 https://codereview.appspot.com/97400043/	Andrew Wilkins	9 years ago

Older »