Elek, Marton created HDDS-2166:
----------------------------------
Summary: Some RPC metrics are missing from SCM prometheus endpoint
Key: HDDS-2166
URL: https://issues.apache.org/jira/browse/HDDS-2166
Project: Hadoop Distributed Data Store
Issue Type: Bug
Reporter: Elek, Marton
Assignee: Elek, Marton
In Hadoop metrics it's possible to register multiple metrics with the same name
but with different tags. For example each RpcServere has an own metrics
instance in SCM.
{code}
"name" :
"Hadoop:service=StorageContainerManager,name=RpcActivityForPort9860",
"name" :
"Hadoop:service=StorageContainerManager,name=RpcActivityForPort9863",
{code}
They are converted by PrometheusSink to a prometheus metric line with proper
name and tags. For example:
{code}
rpc_rpc_queue_time60s_num_ops{port="9860",servername="StorageContainerLocationProtocolService",context="rpc",hostname="72736061cbc5"}
0
{code}
The PrometheusSink uses a Map to cache all the recent values but unfortunately
the key contains only the name (rpc_rpc_queue_time60s_num_ops in our example)
but not the tags (port=...)
For this reason if there are multiple metrics with the same name, only the
first one will be displayed.
As a result in SCM only the metrics of the first RPC server can be exported to
the prometheus endpoint.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org
