JiangHua Zhu created HDDS-10083:
-----------------------------------
Summary: Improve SCM related thread prefix names
Key: HDDS-10083
URL: https://issues.apache.org/jira/browse/HDDS-10083
Project: Apache Ozone
Issue Type: Improvement
Components: SCM
Affects Versions: 1.3.0
Reporter: JiangHua Zhu
When the SCM service is running, many threads use the same prefix name. The
prefix name here directly uses scmUuid. The advantage is that it is easy to
identify.
scmUuid: bc228776-c5ab-5eb5-c35c-3b157cc13381
But a new problem will arise, that is, the thread name is too long.
Here are some basic logs:
{code:java}
2024-01-04 16:05:57,930 [bc228776-c5ab-5eb5-c35c-3b157cc13381-server-thread470]
INFO org.apache.ratis.server.raftlog.segmented.SegmentedRaftLogWorker:
bc228776-c5ab-5eb5-c35c-3b157cc13381@group-5B5552266EB7-SegmentedRaftLogWorker:
Rolling segment log-60086216_60119685 to index:60119685
{code}
{code:java}
2024-01-03 15:04:19,132 [bc228776-c5ab-5eb5-c35c-3b157cc13381-impl-thread1]
INFO org.apache.ratis.server.impl.RoleInfo:
bc228776-c5ab-5eb5-c35c-3b157cc13381: start
bc228776-c5ab-5eb5-c35c-3b157cc13381@group-5B5552266EB7-FollowerState
{code}
The length of scmUuid here reaches 36 bits.
I think the prefix name here should be reduced, such as using ip+port, which
will be much smaller.
for example:
Thread prefix name: bc228776-172.10.10.56-9894
new logs:
{code:java}
2024-01-04 16:05:57,930 [bc228776-172.10.10.56-9894-server-thread470] INFO
org.apache.ratis.server.raftlog.segmented.SegmentedRaftLogWorker:
bc228776-172.10.10.56-9894@group-5B5552266EB7-SegmentedRaftLogWorker: Rolling
segment log-60086216_60119685 to index:60119685
{code}
I think such improvements are beneficial.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
