summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Merge pull request #20582 from ↵v7.518.53Harald Musum2021-12-192-64/+67
|\ | | | | | | | | vespa-engine/balder/refactor-and-speedup-handlevalue - Refactor and speedup MetricParser.handleValue.
| * Make parse method public to enable testing.Henning Baldersheim2021-12-191-1/+1
| |
| * GC unused loggerHenning Baldersheim2021-12-191-2/+0
| |
| * - Refactor and speedup MetricParser.handleValue.Henning Baldersheim2021-12-192-61/+66
|/ | | | | - Only use streaming jackson parsing - Aggregate hash codes to make unique dimension keys.
* Merge pull request #20581 from vespa-engine/balder/reduce-usage-of-hamcrest-2Jon Bratseth2021-12-1921-356/+245
|\ | | | | Simplify testing by sticking to assertEquals/True/False
| * Simplify testing by sticking to assertEquals/True/FalseHenning Baldersheim2021-12-1921-356/+245
|/
* Merge pull request #20580 from vespa-engine/balder/reduce-usage-of-hamcrestJon Bratseth2021-12-1921-248/+205
|\ | | | | Balder/reduce usage of hamcrest
| * Simplify testing by sticking to assertEquals/True/FalseHenning Baldersheim2021-12-1911-85/+56
| |
| * Simnplify testing by sticking to assertEquals/True/FalseHenning Baldersheim2021-12-1912-163/+149
|/
* Merge pull request #20579 from ↵Jon Bratseth2021-12-1810-93/+129
|\ | | | | | | | | vespa-engine/balder/expose-memory-percentage-on-the-container-cluster Balder/expose memory percentage on the container cluster
| * Add a test that heap size percentage override is taken into account when ↵Henning Baldersheim2021-12-184-16/+48
| | | | | | | | allowing memory for proton.
| * If it is a combined cluster take the memory percentage from the ↵Henning Baldersheim2021-12-182-8/+11
| | | | | | | | ContainerCluster as reserved.
| * Let SearchNode and ContentSearchCluster relate to fractionOfMemoryReserved ↵Henning Baldersheim2021-12-183-29/+27
| | | | | | | | instead of a boolean 'combined'
| * Let NodeResourceTuning relate to fractionOfMemoryReserved instead of a ↵Henning Baldersheim2021-12-183-35/+31
| | | | | | | | boolean 'combined'
| * Expose memorypercentage on the ContainerClusterHenning Baldersheim2021-12-182-9/+16
|/
* Merge pull request #20572 from vespa-engine/jonmv/stabilize-threaded-unit-testBjørn Christian Seime2021-12-181-11/+20
|\ | | | | Ensure "retried" has partial result by syncing with result thread bet…
| * Ensure "retried" has partial result by syncing with result thread between ↵Jon Marius Venstad2021-12-171-11/+20
| | | | | | | | retry and doom
* | Merge pull request #20578 from ↵Jon Bratseth2021-12-1820-259/+204
|\ \ | | | | | | | | | | | | vespa-engine/revert-20568-jonmv/replace-optimaize-with-lingua Revert "Replace optimaize with OpenNLP language detector [run-systemtest]"
| * | Revert "Replace optimaize with OpenNLP language detector [run-systemtest]"Jon Marius Venstad2021-12-1820-259/+204
| | |
* | | Merge pull request #20574 from vespa-engine/hakonhall/remove-crash-files-on-hostBjørn Meland2021-12-181-3/+2
|\ \ \ | |/ / |/| | Make UnixPath::deleteIfExists return boolean
| * | Make UnixPath::deleteIfExists return booleanHåkon Hallingstad2021-12-171-3/+2
| | |
* | | Merge pull request #20577 from vespa-engine/hmusum/fix-time-for-last-warningHenning Baldersheim2021-12-181-2/+2
|\ \ \ | | | | | | | | Fix calculation of initial value of last logged timestamp
| * | | Fix calculation of initial value of last logged timestampHarald Musum2021-12-181-2/+2
| | | |
* | | | Merge pull request #20576 from vespa-engine/balder/use-correct-timestampHarald Musum2021-12-187-26/+33
|\ \ \ \ | | | | | | | | | | Use the Instant as a timestamp to make it clear what it is instead of…
| * | | | Use the Instant as a timestamp to make it clear what it is instead of just a ↵Henning Baldersheim2021-12-187-26/+33
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | generic long. Make sure to use it correctly for the system metrics.
* | | | | Merge pull request #20568 from vespa-engine/jonmv/replace-optimaize-with-linguaJon Marius Venstad2021-12-1820-204/+259
|\ \ \ \ \ | |/ / / / |/| | | | Replace optimaize with OpenNLP language detector [run-systemtest]
| * | | | Re-add filesJon Marius Venstad2021-12-185-0/+142
| | | | |
| * | | | Move model to module where it is needed, to simplify, at the cost of larger ↵Jon Marius Venstad2021-12-1815-217/+23
| | | | | | | | | | | | | | | | | | | | bundles
| * | | | Fix dependenciesJon Marius Venstad2021-12-173-0/+18
| | | | |
| * | | | Replace UrlcharSequenceNormalizer with one with an improved regexJon Marius Venstad2021-12-175-6/+80
| | | | |
| * | | | Mockito test scopeJon Marius Venstad2021-12-171-0/+1
| | | | |
| * | | | Add some javadoc, and no need to handle null return for modelJon Marius Venstad2021-12-172-2/+4
| | | | |
| * | | | Avoid guavaJon Marius Venstad2021-12-171-3/+2
| | | | |
| * | | | Replace optimaize with OpenNLP language detectorJon Marius Venstad2021-12-1719-195/+208
| | | | |
* | | | | Merge pull request #20573 from vespa-engine/balder/preserve-number-typeHenning Baldersheim2021-12-1714-181/+213
|\ \ \ \ \ | | | | | | | | | | | | Balder/preserve number type
| * | | | | Add test for dimensions incorrectly containing nullHenning Baldersheim2021-12-171-0/+25
| | | | | |
| * | | | | Handle that metrics can arrive with a dimension set to null. Skip and log it ↵Henning Baldersheim2021-12-173-20/+41
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | as debug. Let exceptions through in the tests so that we do not hide errors there.
| * | | | | Get stacktrace for exceptions.Henning Baldersheim2021-12-171-5/+4
| | | | | |
| * | | | | Drop using hamcrest matchers and stick to simple assertEquals/True/FalseHenning Baldersheim2021-12-179-132/+119
| | | | | |
| * | | | | Preserve number typeHenning Baldersheim2021-12-172-27/+27
| | | | | |
* | | | | | Merge pull request #20575 from vespa-engine/jonmv/remove-stale-notificationsHarald Musum2021-12-171-4/+7
|\ \ \ \ \ \ | |/ / / / / |/| | | | | Remove all stale notifications on submissions
| * | | | | Remove all stale notifications on submissionsJon Marius Venstad2021-12-171-4/+7
| | |_|_|/ | |/| | |
* | | | | Merge pull request #20571 from vespa-engine/balder/consider-consumer-earlierHenning Baldersheim2021-12-174-26/+75
|\ \ \ \ \ | | | | | | | | | | | | Even if we do not filter by consumer on the services, we can do it ea…
| * | | | | Even if we do not filter by consumer on the services, we can do it early in ↵Henning Baldersheim2021-12-174-26/+75
| | |/ / / | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | the metrics-proxy. This is especially benefiscal for the most frequent requests like autoscaling which only looks at a small set (1%) of the metrics. This both reduces memory footprint and saves cpu.
* | | | | Merge pull request #20566 from vespa-engine/hmusum/validate-jvm-options-2Harald Musum2021-12-172-19/+70
|\ \ \ \ \ | |/ / / / |/| | | | Log warning if invalid or unsupported JVM GC options are used
| * | | | Log warning if invalid or unsupported JVM GC options are usedHarald Musum2021-12-172-19/+70
| | | | |
* | | | | Merge pull request #20570 from ↵Tor Brede Vekterli2021-12-173-11/+43
|\ \ \ \ \ | | | | | | | | | | | | | | | | | | | | | | | | vespa-engine/vekterli/fix-additional-update-write-repair-metrics-wiring Complete wiring of OK/failure metric reporting during update write-repair [run-systemtest]
| * | | | | Complete wiring of OK/failure metric reporting during update write-repairTor Brede Vekterli2021-12-173-11/+43
| | |_|_|/ | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This also changes the aggregation semantics of "not found" results. An update reply indicating "not found" is protocol-wise reported back to the client as successful but with a not-found field set that has to be explicitly checked. But on the distributor the `notfound` metric is under the `updates.failures` metric set, i.e. not counted as a success. To avoid double bookkeeping caused by treating it as both OK and not, we choose to not count "not found" updates under the `ok` metric. But to avoid artificially inflating the `failures.sum` metric, we remove the `notfound` metric from the implicit sum aggregation. This means that visibility into "not found" updates requires explicitly tracking the `notfound` metric as opposed to just the aggregate, but this should be an acceptable tradeoff to avoid false failure positives.
* | | | | Merge pull request #20569 from vespa-engine/bratseth/no-notnullJon Marius Venstad2021-12-1746-205/+73
|\ \ \ \ \ | |_|_|_|/ |/| | | | Avoid org.jetbrains annotations
| * | | | Avoid org.jetbrains annotationsJon Bratseth2021-12-1746-205/+73
|/ / / /