aboutsummaryrefslogtreecommitdiffstats
path: root/metrics-proxy/src/test/java/ai
Commit message (Collapse)AuthorAgeFilesLines
* Make system metrics testable.Henning Baldersheim2021-10-141-3/+118
|
* Update Verizon Media copyright notices.gjoranv2021-10-073-3/+3
|
* Do not fetch metrics for default consumer unless explicitly asked to do so.Henning Baldersheim2021-09-212-0/+5
|
* Use an async http client with only 2 threads.Henning Baldersheim2021-09-162-21/+29
|
* Try to fix unstable unit testsHarald Musum2021-09-162-11/+7
| | | | Make taskTimeout volatile
* In order to avoid possible deadlock when calling yourself for metrics when ↵Henning Baldersheim2021-09-153-25/+34
| | | | | | TTL expires, we only fetch metrics in the background. If it is present use it, if not it will be there next time.
* Revert "Revert "Consume and process metrics as they are parsed.""Harald Musum2021-09-093-19/+24
|
* Revert "Consume and process metrics as they are parsed."Harald Musum2021-09-093-24/+19
|
* Consume and process metrics as they are parsed.Henning Baldersheim2021-09-093-19/+24
| | | | This avoids avoid pooling up many metric objects prior to processing and reduces memory pressure.
* Unify on using the MetricId class to avoid creating many instances on the fly.Henning Baldersheim2021-09-029-34/+41
|
* The metrics proxy is very memory hungry.Henning Baldersheim2021-09-012-4/+4
| | | | | | This will avoid creating many identical DimensionIds, ServiceIds. It will also prefer using cheap singleton lists where possible. In addition the code relies less on generated Config code.
* admin/slobrok.0 does not always exist ....... anymore.Henning Baldersheim2021-05-272-3/+3
|
* Upgrade from hc4 to hc5.Henning Baldersheim2021-04-071-2/+3
|
* Add copyright headersJon Bratseth2021-03-183-0/+3
|
* Remove usage of org.jsonBjørn Christian Seime2021-02-026-75/+86
|
* Revert "Remove org.json usage [run-systemtest]"Bjørn Christian Seime2021-02-016-86/+75
|
* Remove usage of org.json:jsonBjørn Christian Seime2021-01-296-75/+86
| | | | | | | | Remove most usage of org.json:json Maven artifact. This library does not have a compatible license. Some usage is still left as it's part of our container-search public API. We'll need a major release to fix that. See https://github.com/vespa-engine/vespa/issues/14762 for more details.
* Remove topleveldispatch remnantsHarald Musum2020-09-252-6/+1
|
* No functional changesJon Bratseth2020-09-188-30/+31
|
* Fixed broken test and fixed namingAmund Bergland Kvalsvik2020-07-201-3/+2
|
* added regex and hostname testAmund Bergland Kvalsvik2020-07-171-1/+30
|
* Add prometheus path to the resource list in the root responsegjoranv2020-07-011-8/+12
|
* Fixed review commentsAmund Bergland Kvalsvik2020-06-291-2/+2
|
* Updated ApplicationMetricsHandler. Removed PrometheusV1HandlerAmund Bergland Kvalsvik2020-06-261-4/+4
|
* Cache built and processed metrics packets instead of buildersgjoranv2020-04-291-4/+4
| | | | .. to avoid concurrent handler requests trying to mutate builders.
* Limit number of threads for applications with <20 nodes.gjoranv2020-04-281-0/+15
|
* Remove assumption of VESPA_HOME location in TelegrafTestOla Aunrønning2020-04-161-1/+2
|
* Reuse Supervisor instanceBjørn Christian Seime2020-03-192-37/+40
| | | | This will allow the client to reuse existing SSL sessions
* use RPC directly instead of calling external processArne Juul2020-03-182-22/+3
|
* memory.util -> mem.utilJon Bratseth2020-03-101-1/+1
|
* Determine metrics API protocol based on isHostedVespaOla Aunrønning2020-02-271-0/+1
|
* Log warning when stdErr is not empty. More debug loggingOla Aunrønning2020-02-211-1/+1
|
* Map input plugins to appropriate output pluginOla Aunrønning2020-02-201-4/+2
|
* Added testOla Aunrønning2020-02-201-0/+44
|
* Add metrics/v2 handlergjoranv2020-02-145-165/+266
| | | | | - Add new config so that hostname and role can be added to output - Refactor tests to avoid duplicating code
* Allow inspecting consumers in MetricsPacket.Buildergjoranv2020-02-131-2/+13
|
* Move metrics handler utilities to new package in container-coregjoranv2020-01-231-25/+0
|
* Implement metrics dimensions processing.gjoranv2020-01-224-0/+184
| | | | | | | | | - retain only public dimensions - clusterId - serviceId + Move constants for public dimensions from config-model to new class in metrics-proxy
* Merge pull request #11840 from vespa-engine/gjoranv/node-infogjoranv2020-01-203-11/+14
|\ | | | | Gjoranv/node info
| * Add node's role to json output.gjoranv2020-01-172-0/+4
| |
| * Expose hostname, and rename nodeId -> role.gjoranv2020-01-173-11/+10
| |
* | Move MetricsHandler to new package and add 'V1' to its name.gjoranv2020-01-172-10/+11
|/
* Enable the default public metrics consumer in hosted Vespa.gjoranv2020-01-141-0/+25
| | | | | - Always filter this consumer from yamas routing, as the effect of including it is uncertain.
* Move node name out of the 'node' object.gjoranv2020-01-092-4/+4
| | | | | - The 'node' object is for node metrics, so it makes more sense to have the node name as a sibling to 'node' and 'services'.
* Rename configId to nodeId in MetricsNodesConfiggjoranv2020-01-062-2/+2
|
* Add/corect copyright headersJon Bratseth2020-01-0332-132/+33
|
* Remove convenience method that did not take consumer argument.gjoranv2020-01-031-5/+5
|
* Properly support multiple consumers when caching metrics.gjoranv2020-01-031-0/+32
|
* Fix a bug in conversion to metrics packetsgjoranv2019-12-261-9/+31
| | | | | - NPE was thrown when there is no 'node' json. - Add testing for this case.
* Rename vespa-nodes.def to metrics-nodes.defgjoranv2019-12-242-9/+9
|