aboutsummaryrefslogtreecommitdiffstats
path: root/configserver
Commit message (Collapse)AuthorAgeFilesLines
* Do not synchronize when bootstrapping tenantsHarald Musum2021-01-261-1/+1
| | | | | This should no longer be necessary, as backing for tenants is a synchronized map
* Merge pull request #16216 from vespa-engine/hmusum/load-sessions-in-parallelHarald Musum2021-01-261-10/+41
|\ | | | | Load sessions in parallel
| * Some fixes to logging after code reviewHarald Musum2021-01-251-16/+15
| |
| * Load sessions in parallelHarald Musum2021-01-251-13/+45
| |
* | Make sure to add the session loaded to cache if it really was loadedHarald Musum2021-01-261-7/+9
| |
* | Revert "Revert "Use synchronized backing for sessions""Harald Musum2021-01-252-27/+33
| |
* | Merge pull request #16214 from ↵Harald Musum2021-01-252-33/+27
|\ \ | | | | | | | | | | | | vespa-engine/revert-16208-hmusum/use-synchronized-backing-for-sessions Revert "Use synchronized backing for sessions"
| * | Revert "Use synchronized backing for sessions"Harald Musum2021-01-252-33/+27
| |/
* / Retry when verifying hostsHarald Musum2021-01-251-3/+22
|/ | | | | | Hosts used might not be in sync on all config servers (we wait for 2/3 servers to respond to deployments and deletions). Retry verification of hosts for 1 second
* Use synchronized collections for cached and watchersHarald Musum2021-01-251-25/+28
| | | | Avoid locking on instance level by using synchronized collections
* Add some more info to logging when creating tenantHarald Musum2021-01-251-2/+5
|
* Make sure to write metadata and add tenant before notifying listenerHarald Musum2021-01-251-2/+2
|
* No need to create and write tenant metadata if tenant already existsHarald Musum2021-01-251-5/+1
|
* Revert "Try to reduce contention when bootstrapping tenants [run-systemtest]"Harald Musum2021-01-258-25/+26
|
* Split out getOrCreateTenant and do not synchronize createTenantHarald Musum2021-01-241-5/+9
| | | | | Split out so that we can avoid the need to synchronized on instance, since we now just do one operation on the synchronized tenants map
* Throw exception if trying to add a tenant that already existsHarald Musum2021-01-248-21/+16
|
* Control which lidspace job implementation to use.Henning Baldersheim2021-01-231-0/+3
|
* Revert "Do not synchronize on method used to bootstrap tenants [run-systemtest]"Harald Musum2021-01-221-6/+3
|
* Do not synchronize on method used to bootstrap tenantsHarald Musum2021-01-221-3/+6
|
* Merge pull request #16152 from ↵Håkon Hallingstad2021-01-2132-282/+87
|\ | | | | | | | | vespa-engine/hmusum/reduce-use-of-global-component-registry-part-3 Reduce use of global component registry, part 3 [run-systemtest]
| * Remove GlobalComponentRegistry and implementationsHarald Musum2021-01-2132-252/+47
| |
| * Move ReloadListener out of GlobalComponentRegistryHarald Musum2021-01-2110-41/+51
| |
* | Merge pull request #16147 from vespa-engine/freva/blocking-prepareValerij Fredriksen2021-01-213-8/+66
|\ \ | |/ |/| Wait for resources in prepare
| * Create new transaction for each checkValerij Fredriksen2021-01-211-2/+2
| |
| * Make it possible to wait for resources in resourcesValerij Fredriksen2021-01-213-8/+66
| |
* | Merge pull request #16143 from ↵Harald Musum2021-01-2133-366/+375
|\ \ | | | | | | | | | | | | vespa-engine/hmusum/reduce-use-of-global-component-registry-part-2 Reduce use of global component registry, part 2 [run-systemtest]
| * | Move ConfigDefinitionRepo out of GlobalComponentRegistryHarald Musum2021-01-2115-64/+45
| | |
| * | Move ModelFactoryRegistry out of GlobalComponentRegistryHarald Musum2021-01-2117-101/+83
| | |
| * | Move Clock out of GlobalComponentRegistryHarald Musum2021-01-2118-55/+57
| | |
| * | Move Zone out of GlobalComponentRegistryHarald Musum2021-01-2112-57/+47
| | |
| * | Move ConfigServerDB out of GlobalComponentRegistryHarald Musum2021-01-2130-91/+101
| | |
| * | Move secret ConfigserverConfig out of GlobalComponentRegistryHarald Musum2021-01-2029-84/+128
| | |
* | | Log time use when building models (debug log)Harald Musum2021-01-211-4/+7
| | |
* | | andreer/endpoint certificate maintainer (#16099)Andreas Eriksen2021-01-216-83/+20
| |/ |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * remove support for old formats and introduce EndpointCertificateMaintainer * record certificate refresh time, run maintainer every 12 hours * retrigger prod deployments if refreshed certificate not deployed after one week * only re-trigger production jobs * unit test EndpointCertificateMaintainer * take application lock to avoid concurrent modifications when managing endpoint certs * only trigger deployment jobs Co-authored-by: Jon Marius Venstad <jonmv@users.noreply.github.com>
* | Merge pull request #16116 from ↵gjoranv2021-01-201-1/+0
|\ \ | | | | | | | | | | | | vespa-engine/gjoranv/do-not-setup-StateMetricConsumerFactory Don't add the StateMetricConsumerFactory component.
| * | Don't add the StateMetricConsumerFactory component.gjoranv2021-01-201-1/+0
| | | | | | | | | | | | | | | - The config model always adds the JdiscMetricsFactory, which should take precendence.
* | | Merge pull request #16107 from vespa-engine/bjorncs/zstd-access-logBjørn Christian Seime2021-01-201-0/+3
|\ \ \ | |_|/ |/| | Bjorncs/zstd access log
| * | Add feature flag for enabling zstd compression of access logsBjørn Christian Seime2021-01-201-0/+3
| |/
* | Move provisioner out of GlobalComponentRegistryHarald Musum2021-01-2015-70/+69
| |
* | Move secret store out of GlobalComponentRegistryHarald Musum2021-01-2011-66/+49
| |
* | Remove zkCacheExecutor from GlobalComponentRegistryHarald Musum2021-01-2010-36/+32
| |
* | Merge branch 'master' into hmusum/reduce-use-of-global-component-registryHarald Musum2021-01-2023-250/+361
|\|
| * Merge pull request #16087 from vespa-engine/hmusum/log-at-lower-levelJon Marius Venstad2021-01-181-3/+3
| |\ | | | | | | Log at lower level
| | * Log at lower levelHarald Musum2021-01-181-3/+3
| | |
| * | Improved model for reindexing dataJon Marius Venstad2021-01-1811-174/+132
| |/ | | | | | | | | | | | | | | | | Making the users of reindexing data config aware allows getting rid of the confusing default-for-app-and-cluster status, and the dummy first-time reindexing in the reindexer. Since the upgraded config servers will wipe the "common" status, the non-upgraded config servers will read these as EPOCH, instead of some time last year. This should not be a problem for any users of these data, as far as I can tell
| * Move logic for indexed document types inside config model api wrappingJon Marius Venstad2021-01-153-39/+7
| |
| * Update another testJon Marius Venstad2021-01-151-9/+0
| |
| * Let all ApplicationReindexing users config-awareJon Marius Venstad2021-01-149-58/+190
| | | | | | | | | | This allows checking for valid clusters and document types when manipulating status. This commit also updates responses in the application/v2/ API to reflect this
| * Let maintainer remove data for no-longer-existent typesJon Marius Venstad2021-01-143-3/+94
| |
| * More cleanupJon Marius Venstad2021-01-145-30/+1
| |