summaryrefslogtreecommitdiffstats
path: root/orchestrator
Commit message (Collapse)AuthorAgeFilesLines
* Revert "Revert "Provide service-monitor and application-model""Håkon Hallingstad2017-10-191-2/+2
|
* Revert "Provide service-monitor and application-model"Håkon Hallingstad2017-10-191-2/+2
|
* Merge pull request #3615 from ↵Håkon Hallingstad2017-10-191-2/+2
|\ | | | | | | | | vespa-engine/hakon/provide-service-monitor-and-application-model Provide service-monitor and application-model
| * Provide service-monitor and application-modelHåkon Hallingstad2017-10-021-2/+2
| |
* | Temporarily ignore unstable orchestrator testBjørn Christian Seime2017-10-031-0/+1
|/
* Make Orchestrator depend on config model apiHåkon Hallingstad2017-10-021-0/+6
|
* Depend on ServiceMonitor interface instead of implHåkon Hallingstad2017-09-291-8/+8
|
* Revert "Revert "Aressem/remove post install script""Arnstein Ressem2017-09-271-0/+2
|
* Revert "Aressem/remove post install script"Arnstein Ressem2017-09-271-2/+0
|
* Remove global install of files and put this in the modules that owns them.Arnstein Ressem2017-09-251-0/+2
|
* Merge pull request #3423 from ↵Bjørn Christian Seime2017-09-182-161/+9
|\ | | | | | | | | vespa-engine/hakon/remove-current-thread-restriction-on-orchestrator-lock Remove current-thread restriction on Orchestrator lock
| * Remove 2Håkon Hallingstad2017-09-181-2/+2
| |
| * Remove current-thread restriction on Orchestrator lockHåkon Hallingstad2017-09-152-161/+9
| |
* | Move orchestrator config to itw own packageHarald Musum2017-09-162-0/+2
| |
* | Revert "Revert "Move orchestrator.def to configdefinititions/""Harald Musum2017-09-112-5/+6
|/
* Revert "Move orchestrator.def to configdefinititions/"Harald Musum2017-09-072-6/+5
|
* Merge pull request #3354 from ↵Harald Musum2017-09-072-5/+6
|\ | | | | | | | | vespa-engine/musum/move-orchestrator-config-definition Move orchestrator.def to configdefinititions/
| * Move orchestrator.def to configdefinititions/Harald Musum2017-09-062-5/+6
| |
* | Merge pull request #3340 from vespa-engine/hakon/add-preferred-orchestrator-lockhakonhall2017-09-062-47/+72
|\ \ | | | | | | Add preferred Orchestrator lock
| * | Add preferred Orchestrator lockHåkon Hallingstad2017-09-062-47/+72
| |/
* / remove only use of autovalueAndreas Eriksen2017-09-062-16/+41
|/
* Remove corp referencesJon Bratseth2017-08-082-3/+3
|
* Reduce output when printing StorageNodeImplHåkon Hallingstad2017-07-111-1/+1
|
* Do not rollback after failed suspendAllHåkon Hallingstad2017-06-152-60/+35
|
* Update copyright headersJon Bratseth2017-06-1472-68/+72
|
* Revert "Update copyright headers"Jon Bratseth2017-06-1472-72/+68
|
* Update copyright headersJon Bratseth2017-06-1472-68/+72
|
* Remove carriage returnJon Bratseth2017-06-143-3/+3
|
* Revert "Copyright header"Jon Bratseth2017-06-1372-75/+71
|
* Copyright headerJon Bratseth2017-06-1372-71/+75
|
* Make patch workHåkon Hallingstad2017-06-131-2/+1
| | | | | | Removes the PathParams annotation on the subclass, which caused Jersey to ignore the annotations on the superclass(!?). I have verified these changes work in CD.
* Merge pull request #2684 from yahoo/mpolden/remove-hostname-referencesValerij Fredriksen2017-06-122-12/+12
|\ | | | | Remove hostname references
| * Remove hostname references in orchestrator moduleMartin Polden2017-06-092-12/+12
| |
* | Orchestrator support for setting host statusHåkon Hallingstad2017-06-125-1/+104
|/ | | | | | | | | | | | | | | | Directly setting the host status can be useful for an operator, e.g. to break a deadlock. Unfortunately, some code use the term "host status" while others use "host state". I haven't settled yet on which is preferred, e.g. 'state' is used in REST APIs, but 'status' is the original term and slightly more used. This PR maintains the local use of the terms, meaning it adds 'state' where that's normally used and 'status' other places. Setting the deadlock is done with a PATCH request, which is defined in jaxrs_utils. jaxrs_utils only defines PATCH, and that's not used anywhere else except in jaxrs_client_utils tests. So I remove this module as a dependency a couple of places.
* Complete retirement early in dev CDHåkon Hallingstad2017-05-2623-773/+348
|
* Merge pull request #2431 from yahoo/hakon/add-path-at-application-lock-timeoutHarald Musum2017-05-101-1/+1
|\ | | | | Add path at application lock timeout
| * Add path at application lock timeoutHåkon Hallingstad2017-05-101-1/+1
| |
* | Merge pull request #2378 from yahoo/hakonhall/use-orchestrator-modelhakonhall2017-05-0814-302/+377
|\ \ | |/ |/| Makes the Orchestrator policy use the new model classes
| * Makes the Orchestrator policy use the new model classesHåkon Hallingstad2017-05-0314-302/+377
| | | | | | | | | | | | | | | | | | | | Functionally, the new code should be identical to the old one, EXCEPT if an application has 2 or more nodes on a Docker host: - In this case the new code puts these nodes into a NodeGroup, and will try to suspend them all at the same time. This PR could have been functionally identical to the old code by by putting each HostName into its own NodeGroup in OrchestratorImpl::nodeGroupsOrderedForSuspend.
* | Make TestUtil generate correct config ID for cluster controller, take 2Håkon Hallingstad2017-05-041-1/+1
| |
* | Make TestUtil generate correct config ID for cluster controllerHåkon Hallingstad2017-05-041-1/+1
| |
* | Correctly extract cluster controller index from standalone cluster controllerHåkon Hallingstad2017-05-042-4/+14
| |
* | Remove redundant testHarald Musum2017-05-041-6/+0
| |
* | Fix getting cluster controller index for standalone clustersHarald Musum2017-05-044-11/+23
|/ | | | | | | | * Cluster controller cluster can either be using zookeeper on config server and be part of admin cluster (config id 'admin/cluster-controllers/[n]') or standalone (running zookeeper in cluster controller and be in its own cluster, config id '[content cluster id]/standalone/[n]') VESPA-7356
* Adds classes to give the Orchestrator policy classes a simplified view of Vespa.Håkon Hallingstad2017-04-2824-70/+1226
| | | | | | | | | This should be a no-op. The only changes that actually could have an impact are the changes to getting the cluster controllers, but it should be functionally equivalent. This PR will make it easier to change the Orchestrator policy to allow suspending several nodes (NodeGroup) in an application on a single Docker host.
* newHashSet is not recommended to use, triggers warningsArne H Juul2017-04-271-2/+2
|
* avoid rawtypes warningsArne H Juul2017-04-271-1/+1
| | | | | * the warning here is due to a JDK bug, but the workaround is simple.
* Add node-admin suspendvalerijf2017-04-031-25/+19
|
* Remove old legacyJon Bratseth2017-03-281-1/+1
|
* Nonfunctional changesJon Bratseth2017-03-211-1/+1
|