summaryrefslogtreecommitdiffstats
path: root/configserver
Commit message (Collapse)AuthorAgeFilesLines
* Purge old sessions every 5 minutes and log when doing soHarald Musum2018-01-221-1/+3
|
* Refactor prepare handlerHarald Musum2018-01-217-62/+110
| | | | Move more stuff into business layer, no functional changes
* Refer to generated tensors by absolute pathJon Bratseth2018-01-191-2/+1
|
* Always write tensor binary files when possibleJon Bratseth2018-01-192-2/+4
| | | | | | The application package is read in two locations, the uploaded location and the ".preprocessed" location. We need to produce tensor binaries each time, even though the ranking constants seemlingly survives from one to the other.
* Use a constant for error codeHarald Musum2018-01-181-1/+2
|
* Do not log connection errors (services may not be up, common case)Harald Musum2018-01-181-1/+1
|
* We might have status unknown for all hosts tooHarald Musum2018-01-173-7/+44
|
* Merge pull request #4692 from ↵Jon Bratseth2018-01-172-14/+23
|\ | | | | | | | | vespa-engine/bratseth/store-converted-expressions-in-zk Bratseth/store converted expressions in zk
| * Persist constant info in ZooKeeperJon Bratseth2018-01-171-2/+2
| |
| * Copy generated files to ZooKeeperJon Bratseth2018-01-171-8/+17
| |
| * Access files through application packageJon Bratseth2018-01-161-6/+6
| |
* | Merge pull request #4672 from ↵Henning Baldersheim2018-01-163-7/+21
|\ \ | | | | | | | | | | | | vespa-engine/hmusum/ask-config-servers-to-download-files Ask other config servers to start downloading files
| * | Ask other config servers to start downloading filesHarald Musum2018-01-163-7/+21
| | | | | | | | | | | | | | | | | | To have redundancy, ask the other config servers to start downloading a file, so we have redundancy in case the config server running the code goes down after deployment has finished
* | | Merge pull request #4665 from ↵Harald Musum2018-01-163-31/+25
|\ \ \ | |/ / |/| | | | | | | | vespa-engine/hmusum/purge-old-sessions-as-a-maintenance-task Purge old session in its own thread instead of every time we add a se…
| * | SimplifyHarald Musum2018-01-151-5/+0
| | |
| * | Purge old session in its own thread instead of every time we add a sessionHarald Musum2018-01-153-26/+25
| | |
* | | Merge pull request #4661 from ↵Harald Musum2018-01-151-0/+3
|\ \ \ | |/ / |/| | | | | | | | vespa-engine/hmusum/include-configserver-config-file Make it possible to include config server config in a file
| * | Make it possible to include config server config in a fileHarald Musum2018-01-151-0/+3
| | |
* | | Use vespamallocd to better root out jni memory issues.Henning Baldersheim2018-01-151-0/+2
|/ /
* | Reduce log level when not finding fileHarald Musum2018-01-151-3/+2
| | | | | | | | Also make method void, since return value was never used
* | Make sure to cleanup everything when deleting a tenantHarald Musum2018-01-154-32/+28
|/ | | | | * Simplify and make sure to always call close(), return correct exception when tenant is not found when deleting
* Log when before and after creating tenantsHarald Musum2018-01-141-5/+6
|
* Merge pull request #4633 from ↵Harald Musum2018-01-132-2/+2
|\ | | | | | | | | vespa-engine/hakonhall/some-curator-clients-require-ensemble-connect-string Some Curator clients require ensemble connect string
| * Some Curator clients require ensemble connect stringHåkon Hallingstad2018-01-112-2/+2
| |
* | Do not try to access RPC return values when we get an errorHarald Musum2018-01-121-17/+18
|/ | | | | Return values might not exist when we get an error from an RPC reques, return or throw instead.
* Merge pull request #4617 from vespa-engine/hmusum/always-delete-temp-dirValerij Fredriksen2018-01-111-1/+1
|\ | | | | Always delete temp dir
| * Always delete temp dirHarald Musum2018-01-111-1/+1
| |
* | Use constant tensor files WIPJon Bratseth2018-01-114-10/+11
|/
* CHeck file distribution status for hosts in parallelHarald Musum2018-01-111-3/+24
|
* Use higher precision for timeoutHarald Musum2018-01-091-1/+1
|
* Support milliscond resolution for timeout request parameterHarald Musum2018-01-092-5/+13
|
* Timeout for file distribution status APIHarald Musum2018-01-096-34/+33
| | | | | * Support timeout request parameter * Change default timeout to 5 seconds
* Add file distribution status APIHarald Musum2018-01-097-10/+374
|
* Throw ActivationConflictException when unable to acquire activate lockHarald Musum2018-01-031-1/+1
|
* Revert "Revert "Revert "Revert "Jvenstad/zone cleanup 4""""Jon Marius Venstad2018-01-031-1/+1
| | | | This reverts commit 7b99d259bf378c0e5c43c9a199efffdf056678f4.
* Don't test for exact version, just that we get oneHarald Musum2018-01-031-1/+1
|
* Ad some javadoc and do some minor tweaksHarald Musum2018-01-032-21/+16
|
* No need to override methods that the superclass does correctlyHarald Musum2018-01-021-9/+0
|
* Rewrite status handlerHarald Musum2018-01-028-201/+112
| | | | | | Rewrite and simplify to use Slime and a regular handler instead of Jersey, so it's not necessary to manually update handler when config changes (which had not been done in years)
* Revert "Jvenstad/zone cleanup 4"Jon Marius Venstad2018-01-021-1/+1
|
* Merge branch 'master' into jvenstad/zone-cleanup-4Jon Marius Venstad2018-01-0235-127/+147
|\
| * use new LoggingRequestHandler constructorArne Juul2017-12-2731-121/+131
| |
| * Remove container-clustercontroller from list of services to checkHarald Musum2017-12-191-2/+0
| | | | | | | | | | container loads the same bundles, so it should be safe to remove clustercontroller completely
| * Skip checking config convergence for clustercontrollerHarald Musum2017-12-191-1/+2
| |
| * Use one supervisor and transport and shut it down once it not needed anymore.Henning Baldersheim2017-12-183-5/+16
| |
* | ConflictsJon Marius Venstad2017-12-155-18/+19
|\|
| * Do async file distribution (when preparing application) when file ↵Harald Musum2017-12-145-18/+19
| | | | | | | | | | | | distributor disable Also: Don't cancel future when getting dile references started by async file distribution
* | Conflict resolvedJon Marius Venstad2017-12-147-15/+69
|\|
| * Set timeout for internal deployment in one placeHarald Musum2017-12-133-3/+19
| | | | | | | | | | Use barrier timeout as basis for timeout, in the same way we do for external deployments.
| * Notify config proxy about starting downloads while preparingHarald Musum2017-12-134-12/+50
| | | | | | | | Code not activated yet