Qlever introduced Prometheus metrics export with MR 2817 (Demo Dashboard).
We should make use of these metrics. Enabling Kubernetes to fetch the metrics is as easy as adding the corresponding port with a name ending in -metrics.
AC:
| Subject | Author | Repo | Branch | Lines +/- | |
|---|---|---|---|---|---|
| WDQS: Enable prometheus metrics on Qlever | Trueg | operations/deployment-charts | master | +2 -1 |
Example:
$ curl localhost:7001/metrics # HELP qlever_index_rebuild_in_progress Whether an index rebuild is currently in progress (1) or not (0) # TYPE qlever_index_rebuild_in_progress gauge qlever_index_rebuild_in_progress 0 # HELP qlever_server_start_time_seconds Unix timestamp when the QLever server was started # TYPE qlever_server_start_time_seconds gauge qlever_server_start_time_seconds 1785233567 # HELP qlever_index_load_time_seconds Unix timestamp when the current index was loaded # TYPE qlever_index_load_time_seconds gauge qlever_index_load_time_seconds 1785233567 # HELP qlever_delta_triples Number of triples that are updated relative to the index # TYPE qlever_delta_triples gauge qlever_delta_triples 0 # HELP qlever_sparql_operation_running Number of SPARQL operations currently being processed # TYPE qlever_sparql_operation_running gauge qlever_sparql_operation_running{operation="query"} 0 qlever_sparql_operation_running{operation="update"} 0 # HELP qlever_memory_cache_used_bytes Memory used for caching # TYPE qlever_memory_cache_used_bytes gauge qlever_memory_cache_used_bytes 0 # HELP qlever_sparql_operation_finished_total Number of SPARQL operations successfully finished since server start # TYPE qlever_sparql_operation_finished_total counter qlever_sparql_operation_finished_total{operation="query"} 0 qlever_sparql_operation_finished_total{operation="update"} 0 # HELP qlever_memory_query_available_bytes Available memory for query processing # TYPE qlever_memory_query_available_bytes gauge qlever_memory_query_available_bytes 120000000000 # HELP qlever_sparql_operation_started_total Number of SPARQL operations started since server start # TYPE qlever_sparql_operation_started_total counter qlever_sparql_operation_started_total{operation="query"} 0 qlever_sparql_operation_started_total{operation="update"} 0 # HELP qlever_http_errors_total Errors during the execution of non SPARQL operations # TYPE qlever_http_errors_total counter qlever_http_errors_total{type="internal"} 0 qlever_http_errors_total{type="http"} 0 # HELP qlever_sparql_operation_errors_total Errors during the execution of SPARQL operations # TYPE qlever_sparql_operation_errors_total counter qlever_sparql_operation_errors_total{type="system_error"} 0 qlever_sparql_operation_errors_total{type="internal"} 0 qlever_sparql_operation_errors_total{type="syntax"} 0 qlever_sparql_operation_errors_total{type="timeout"} 0 qlever_sparql_operation_errors_total{type="send_streamable_response"} 0 qlever_sparql_operation_errors_total{type="protocol"} 0 qlever_sparql_operation_errors_total{type="in_use"} 0 # HELP qlever_memory_query_limit_bytes Memory allocated for query processing # TYPE qlever_memory_query_limit_bytes gauge qlever_memory_query_limit_bytes 120000000000 # HELP qlever_memory_cache_limit_bytes Memory allocated for caching # TYPE qlever_memory_cache_limit_bytes gauge qlever_memory_cache_limit_bytes 32000000000
Change #1321948 had a related patch set uploaded (by Trueg; author: Trueg):
[operations/deployment-charts@master] WDQS: Enable prometheus metrics on Qlever
Qlever metrics are now exposed and tracked in the WIP WDQS grafana board. The Qlever endpoint does not expose all metrics mentioned in the AC though.
Change #1321948 merged by jenkins-bot:
[operations/deployment-charts@master] WDQS: Enable prometheus metrics on Qlever