~dave-cheney/juju-core/124-worker-peergrouper-mock-fix : changes

~dave-cheney/juju-core/124-worker-peergrouper-mock-fix » Changes from revision 2566

From Revision 2566 to 2547

Rev	Summary	Authors	Date
2566	Mock test should order results Mock test should order results	Dave Cheney	10 years ago
2565	[r=axwalk] provider/azure: ensure tests work correctly [r=axwalk] provider/azure: ensure tests work correctly Some tests I wrote call StartInstance to check the parameters passed to createInstance. The tests don't actually verify that it called that method successfully, and all validation is done within the method. Updated the tests to ensure from outside that the method is called. https://codereview.appspot.com/84290044/	Andrew Wilkins	10 years ago
2564	[r=thumper] Have the apiserver know the logdir [r=thumper] Have the apiserver know the logdir The agent config now knows the logdir. In some places the entire agent config is passed through, but in order to keep the changes small(ish), we only pass through the logdir (additionally). This did require a few tweaks to tests, and in particular the addition of LogDir to the base ConnSuite. This logdir is used by the upcoming debug-log api. https://codereview.appspot.com/84350043/	Tim Penhey	10 years ago
2563	[r=thumper] Extend the tailer utility. [r=thumper] Extend the tailer utility. When looking at the pyjuju debug-log, there was a replay function that allowed getting the log from the start. There was no way to tell the tailer to start from the beginning of the file. I changed the meaning of NewTailer to be one that doesn't do any seeking. The NewTailerBacktrack does seek to the end, and then goes back a certain number of lines. There was a bug in the backtrack tailer when zero lines were passed in. This is fixed and a test added. The number of lines to look back is now a uint, as negative values make no sense at all. https://codereview.appspot.com/84290045/	Tim Penhey	10 years ago
2562	[r=wallyworld],[bug=1301464] Port changes from 1.18 which ad... [r=wallyworld],[bug=1301464] Port changes from 1.18 which address bug 1301464 Change the internals of the megawatcher to ensure that machine addresses are correct and that the public and private addresses for untis come from the assigned machine.	Ian Booth	10 years ago
2561	[r=natefinch],[bug=1285410] Fix architecture name for armhf. [r=natefinch],[bug=1285410] Fix architecture name for armhf. juju-tools for arm 32-bit all use "armhf" as the arch name; juju-core should use it too, rather than "arm".	Jason Hobbs	10 years ago
2560	[r=cmars] Add series: attribute to charm metadata. [r=cmars] Add series: attribute to charm metadata. Allows charms to declare a preferred series. The field is optional, but if set, it must be a valid series identifier, using the same criteria as a charm URL.	Casey Marshall	10 years ago
2559	[r=cmars],[bug=1290824] Resolve series with charm store in j... [r=cmars],[bug=1290824] Resolve series with charm store in juju client. For the deploy and upgradecharm commands, when a series is not provided, and a default-series is not set in the environment config, the client will resolve the series with the charm store, through the state server. If the client is working with a 1.16 state server, it will resolve the series directly with the charm store. Existing environments will have a default-series, so they should have no change in series selection. New environments will continue to support the default-series config setting, but it can be omitted, and is not set by default. https://codereview.appspot.com/80280043/	Casey Marshall	10 years ago
2558	[r=fwereade] testing/filetesting: new package [r=fwereade] testing/filetesting: new package ...extracted from utils/zip, and given some actual tests, which (surprise surprise) made a few problems apparent. https://codereview.appspot.com/80540043/	William Reade	10 years ago
2557	[r=axwalk] state: no pre-checks for injected machines [r=axwalk] state: no pre-checks for injected machines We should not be second guessing the bootstrap agent or manual provisioning. If the instance-id is non-empty when adding a machine to state, do not check SupportsUnitPlacement. https://codereview.appspot.com/83940043/	Andrew Wilkins	10 years ago
2556	[r=dave-cheney],[bug=1301719] Do not strip(1) jujud [r=dave-cheney],[bug=1301719] Do not strip(1) jujud Also, while I'm here make sure that if the tools are built with gccgo they are built using a static libgo so the tools will be self contained. https://codereview.appspot.com/83830044/	Dave Cheney	10 years ago
2555	[r=axwalk] provider/azure: availability-sets-enabled config [r=axwalk] provider/azure: availability-sets-enabled config A new config attribute availability-sets-enabled is defined, which controls the mode in which the Azure environment operates. If true (the default), availability set support is enabled and instances will be grouped by service. The attribute cannot be changed after the environment is prepared. If the availability-sets-enabled is true, unit placement is prohibited. This is necessary to avoid various issues such as incorrectly load- balancing ports across heterogenous services, and makes it possible to unambiguously determine which Cloud Service to assign units to for high availability. https://codereview.appspot.com/77950045/	Andrew Wilkins	10 years ago
2554	[r=axwalk] provider/azure: enable load-balancing/availabilit... [r=axwalk] provider/azure: enable load-balancing/availability This change is required to enable Availability Set support for Azure. The existing implementation was fundamentally incompatible with Availability Sets, so the implementation has been significantly altered, while maintaining backwards compatibility. In the old implementation, instances are associated one-to-one with Cloud Services. In the new implementation, instances are associated with Roles within a Cloud Service; there may be multiple Roles within a single Cloud Service. If a service is exposed, then its ports will be load- balanced across each instance within the cloud service which opens that port. We currently open ports 22 for all instances, and the Mongo and API server ports for state server instances. State server instances are allocated to a single Cloud Service. There are a number of incidental improvements, e.g. using an updated version of DeleteHostedService which destroys associated disks/blobs server side. All instances are now allocated to an Availability Set called "juju", which is scoped to the Cloud Service. We do not currently group units into Cloud Services automatically, however the ground work is laid to do this. A follow-up will implement this work. When we enable grouping, the same time we will have to make changes to how we connect to machines via SSH, either by proxying through an arbitrary machine in the Cloud Service, or by not load balancing port 22 and connecting to the machine's auto-allocated public port. https://codereview.appspot.com/73910043/	Andrew Wilkins	10 years ago
2553	[r=dave-cheney],[bug=1300032] fix 1300032: charm: gccgo test... [r=dave-cheney],[bug=1300032] fix 1300032: charm: gccgo test failure When network is down, http-proxy returns an error in html format, hence the leading '<'. Ensure the response code is 200, otherwise return an error.	Jesse Meek	10 years ago
2552	[r=axwalk] Add EnvironCapability to state.Policy [r=axwalk] Add EnvironCapability to state.Policy environs.EnvironCapability is moved to the state package, extended with the SupportsUnitPlacement method, and added to the state.Policy interface. Methods for creating machines and assigning units to them are now modified to use the SupportsUnitPlacement method to enable policy implementations to block adding of empty machines (juju add-machine) and to block unit placement (deploy/add-unit --to). Also, a new EnvironBase type is added to provider/common, which implements some default behaviour common to all providers such as the new SupportsUnitPlacement method. A followup will override the default behaviour of SupportsUnitPlacement in the Azure provider, blocking placement if in availability-sets-enabled mode. https://codereview.appspot.com/77820044/	Andrew Wilkins	10 years ago
2551	[r=axwalk] Add DistributionGroup to StartInstanceParams [r=axwalk] Add DistributionGroup to StartInstanceParams DistributionGroup returns instances that StartInstance should consider when provisioning for high availability (i.e. the new instance + those in DistributionGroup should be distributed for high availability.) DistributionGroup is implemented in terms of a new provisioner API method of the same name. This API returns different instances depending on the machine passed in. For an environ manager, all other environ manager instances are returned; for a non-environ manager, all other instances with services in common are returned. This is in preparation for modifying the Azure provider to group instances into the same cloud service when in availability-sets-enabled mode. https://codereview.appspot.com/75250043/	Andrew Wilkins	10 years ago
2550	[r=dave-cheney],[bug=1251076] Workaround LP 1251076 [r=dave-cheney],[bug=1251076] Workaround LP 1251076 gccgo appears to get confused and thinks that out[1] is not nil when in fact it is an interface value of type error and value nil. This workaround solves the problem by leaving error as nil if in fact it was nil and causes no harm for gc because the predicates above assert that if out[1] is not nil, then it is an error type. https://codereview.appspot.com/83340045/	Dave Cheney	10 years ago
2549	[r=wwitzel3] worker/peergrouper: use mongo.SelectPeerHostPor... [r=wwitzel3] worker/peergrouper: use mongo.SelectPeerHostPort This will ensure that we are using the same method for getting the peer hostPort even if the implementation details of SelectInternalHostPort change. https://codereview.appspot.com/83130045/	Wayne Witzel III	10 years ago
2548	[r=wwitzel3] state/api/agent: added IsMaster to the API [r=wwitzel3] state/api/agent: added IsMaster to the API IsMaster provides a way to let an API worker know that it is on the same machine as the API server that has a mongo peer address that is the primary mongo server for a given replicaset. https://codereview.appspot.com/83150043/	Wayne Witzel III	10 years ago
2547	[r=rogpeppe] worker/singular: new package [r=rogpeppe] worker/singular: new package This acts as a wrapper that will allow us to start workers that require only one instance. https://codereview.appspot.com/83300047/	Roger Peppe	10 years ago

Older »