summaryrefslogtreecommitdiffstats
path: root/jdisc_core
Commit message (Collapse)AuthorAgeFilesLines
* Reduce logging of false positives in container watchdogBjørn Christian Seime2018-03-082-28/+33
| | | | | | Introduce a lower grace period only used by the gc check to ensure that deactivated containers have a chance of being collected before being reported as stale in the log and the metrics api.
* Only run gc when number of stale containers above thresholdBjørn Christian Seime2018-02-062-10/+23
| | | | | | Changes the deactivated container watchdog to only trigger gc when the number of deactivated containers passed grace period is above a threshold.
* Log when activate times out (low-level timeout in jdisc)Harald Musum2018-01-311-1/+1
|
* Revert "Revert "Redefine task and its context""Håkon Hallingstad2018-01-231-0/+8
|
* Revert "Redefine task and its context"Håkon Hallingstad2018-01-231-8/+0
|
* Redefine task and its contextHåkon Hallingstad2018-01-221-0/+8
| | | | | | | | | | | | | After discussions with bratseth, we'll not organize tasks in a tree structure, and instead have tasks mostly 1:1 with components. Tasks are put in chains, and may have ordering constraints between them that the chain resolves. The FileSystem API has been removed and instead use the raw java NIO FileSystem, and use Jimfs to mock FileSystem in tests. NodeAdminStateUpdater has been made an interface, since the HTTP request handler needs to delegate to it when getting requests. Much of the public HTTP API will get internalized with standalone NodeAdmin.
* Use protobuf 3.4.0 by defaultJon Bratseth2018-01-181-1/+1
|
* Nonfunctional changes onlyJon Bratseth2018-01-158-30/+30
|
* remove unnecessary loggingHåkon Hallingstad2017-12-231-1/+0
|
* Move initial log to right after Vespa logging has been initializedHåkon Hallingstad2017-12-231-1/+2
|
* Revert "Revert "Log at JDisc process start and stop""Håkon Hallingstad2017-12-231-1/+4
|
* Revert "Log at JDisc process start and stop"Arne H Juul2017-12-231-4/+1
|
* Log at JDisc process start and stopHåkon Hallingstad2017-12-221-1/+4
|
* Error should also exit JDisc containerHåkon Hallingstad2017-12-081-1/+1
|
* Split parent + container-dependency-versions from root pom.gjoranv2017-12-011-0/+1
| | | | | | - Add missing dependencies so that all provided non-yahoo jars are listed in container-dependency-versions. - Add relativePath for all child poms of parent.
* Revert "Gjoranv/split parent2"gjoranv2017-11-301-1/+0
|
* Split parent + container-dependency-versions from root pom.gjoranv2017-11-301-0/+1
| | | | | | - Add missing dependencies so that all provided non-yahoo jars are listed in container-dependency-versions. - Add relativePath for all child poms of parent.
* Revert "Gjoranv/split parent"gjoranv2017-11-291-1/+0
|
* Split parent + container-dependency-versions from root pom.gjoranv2017-11-291-0/+1
| | | | | | - Add missing dependencies so that all provided non-yahoo jars are listed in container-dependency-versions. - Add relativePath for all child poms of parent.
* Print exception since logging may not have been setupHåkon Hallingstad2017-11-141-0/+2
|
* Merge pull request #4095 from vespa-engine/arnej/add-jdisc-handler-metricsgjoranv2017-11-131-2/+12
|\ | | | | Arnej/add jdisc handler metrics
| * add timeElapsed convenience methodArne Juul2017-11-131-2/+12
| | | | | | | | * also, fix some misleading comments
* | Reduce log noise from HttpServerConformanceTestBjørn Christian Seime2017-11-131-3/+6
|/
* Revert "Bjorncs/jdisc http service cleanup"Martin Polden2017-11-101-6/+3
|
* Merge pull request #4072 from vespa-engine/bjorncs/jdisc-http-service-cleanupgjoranv2017-11-101-3/+6
|\ | | | | Bjorncs/jdisc http service cleanup
| * Reduce log noise from HttpServerConformanceTestBjørn Christian Seime2017-11-091-3/+6
| |
* | mark with deprecatedArne Juul2017-11-091-0/+2
| |
* | add the pattern to BindingMatchArne Juul2017-11-094-10/+40
|/
* Nonfunctional changes onlyJon Bratseth2017-11-081-2/+3
|
* Nonfunctional changes onlyJon Bratseth2017-10-2728-73/+86
|
* Add copyright headerJon Bratseth2017-10-271-0/+1
|
* Revert "Revert "Aressem/remove post install script""Arnstein Ressem2017-09-271-0/+4
|
* Revert "Aressem/remove post install script"Arnstein Ressem2017-09-271-4/+0
|
* Remove global install of files and put this in the modules that owns them.Arnstein Ressem2017-09-251-0/+4
|
* use HostName.getLocalhostArne Juul2017-09-081-7/+2
|
* Add comment about xml-apis dependency.gjoranv2017-09-081-0/+1
| | | | - Comment from old Vespa repo git log.
* Revert "Gjoranv/remove fat model import"gjoranv2017-09-071-1/+0
|
* Add comment explaining why jdisc_core depends on xml-apis.gjoranv2017-09-071-0/+1
| | | | - Comment is from the old vespa repo git log.
* Don't install jdisck logfmt variantJon Bratseth2017-09-051-0/+2
|
* Trigger GC more oftenBjørn Christian Seime2017-08-251-1/+1
|
* Merge pull request #3202 from vespa-engine/gjoranv/severe-loggjoranv2017-08-241-1/+1
|\ | | | | Log exception from StandaloneMain as SEVERE.
| * Log exception from StandaloneMain as SEVERE.gjoranv2017-08-241-1/+1
| |
* | Perform best-effort cleanup to stop JDisc from never closing connectionBjørn Christian Seime2017-08-231-5/+23
|/
* Reduce number of false positivesBjørn Christian Seime2017-08-151-1/+2
| | | | | | Increase grace period of stale containers as there is no guarantee when they are collected. Trigger finalization in case objects with finalizer are (indirectly) referring to ActiveContainer instances.
* Triggering finalization does not guarantee that phantom references are enqueuedBjørn Christian Seime2017-07-202-2/+2
|
* Merge pull request #2966 from yahoo/bjorncs/activecontainer-watchdogBjørn Christian Seime2017-07-171-0/+2
|\ | | | | Temporarily ignore test
| * Temporarily ignore testBjørn Christian Seime2017-07-171-0/+2
| |
* | Improve ActiveContainerDeactivationWatchdog unit testBjørn Christian Seime2017-07-172-7/+17
| |
* | Log when scheduled tasks are triggeredBjørn Christian Seime2017-07-171-0/+3
| |
* | Don't run jobs in parallelBjørn Christian Seime2017-07-171-1/+1
|/ | | | | Running them in parallel is not required, and it is easier to debug/trace issues later if jobs are executed sequentially.