search

apache / accumulo

Apache Accumulo
https://accumulo.apache.org
Apache License 2.0
1.07k stars 445 forks source link

Compaction metrics have redundant queue.id tag #5003

Closed keith-turner closed 3 weeks ago

keith-turner commented 1 month ago

Describe the bug

Compaction metrics have queue.id and resource.group tags that have the same value. Below are metrics from running 4.0.0-SNAPSHOT.

2024-10-18T23:58:19,391 [accumulo.METRICS] INFO : accumulo.compactor.queue.count{host=localhost,instance.name=uno,port=9999,process.name=manager,resource.group=default} value=1
2024-10-18T23:58:19,394 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.avg.age{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=1.039
2024-10-18T23:58:19,394 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.dequeued{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=33
2024-10-18T23:58:19,394 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.max.age{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=1
2024-10-18T23:58:19,395 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.min.age{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=1
2024-10-18T23:58:19,395 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.priority{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=-32747
2024-10-18T23:58:19,395 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.queued{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=17
2024-10-18T23:58:19,395 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.rejected{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=0
2024-10-18T23:58:19,395 [accumulo.METRICS] INFO : accumulo.compactor.queue.length{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} value=10000
2024-10-18T23:58:19,402 [accumulo.METRICS] INFO : accumulo.compactor.queue.jobs.exit.time{host=localhost,instance.name=uno,port=9999,process.name=manager,queue.id=default,resource.group=default} throughput=0.416667/s mean=7.188792874s max=16.985985063s

Expected behavior

The queue.id tag can be dropped in favor of the new resource.group tag.

keith-turner commented 3 weeks ago

Discussed this with @dlmarion and @ddanielr . These may not actually be redundant. The resource.group tag is for the manager process and will always be default ATM. The queue.id tag is per compactor resource group in the compaction coordinator and can differ.