[
https://issues.apache.org/jira/browse/HBASE-24745?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17160883#comment-17160883
]
wenfeiyi666 commented on HBASE-24745:
-------------------------------------
A simple mock test in branch-2.3 and master, everything is normal.
{code:java}
final long initPauseTime = 1000;
int tries = 0;
long pauseTime;
while (tries < 10000) {
try {
throw new IllegalArgumentException();
} catch (Exception e) {
pauseTime = ConnectionUtils.getPauseTime(initPauseTime, tries);
LOG.info("trie=" + tries + ", pauseTime=" + pauseTime);
Threads.sleep(pauseTime);
tries++;
}
}
{code}
output
{code:java}
2020-07-20 12:18:22,331 INFO regionserver.Test(22): trie=0, pauseTime=1001
2020-07-20 12:18:23,342 INFO regionserver.Test(22): trie=1, pauseTime=2004
2020-07-20 12:18:25,348 INFO regionserver.Test(22): trie=2, pauseTime=3011
2020-07-20 12:18:28,363 INFO regionserver.Test(22): trie=3, pauseTime=5002
2020-07-20 12:18:33,368 INFO regionserver.Test(22): trie=4, pauseTime=10019
2020-07-20 12:18:43,389 INFO regionserver.Test(22): trie=5, pauseTime=20051
2020-07-20 12:19:03,444 INFO regionserver.Test(22): trie=6, pauseTime=40300
2020-07-20 12:19:43,751 INFO regionserver.Test(22): trie=7, pauseTime=100958
2020-07-20 12:21:24,717 INFO regionserver.Test(22): trie=8, pauseTime=100961
...{code}
> 'Failed report transition' logs too often
> -----------------------------------------
>
> Key: HBASE-24745
> URL: https://issues.apache.org/jira/browse/HBASE-24745
> Project: HBase
> Issue Type: Sub-task
> Affects Versions: 2.3.0
> Reporter: Michael Stack
> Assignee: wenfeiyi666
> Priority: Minor
>
> The parent issue fixed a backoff that was too aggressive. Now I notice we try
> too much. Saw 9k logs in 17 seconds of the below type...
> {code:java}
> 2020-07-15 14:36:23,104 INFO
> org.apache.hadoop.hbase.regionserver.HRegionServer: Failed report transition
> server { host_name: "X.example.org" port: 16020 start_code: 1594823099666 }
> transition { transition_ code: CLOSED region_info { region_id:
> 1594814749475 table_name { namespace: "default" qualifier:
> "IntegrationTestBigLinkedList" } start_key: "\"\"\"\"\"\"\" " end_key:
> "#Q\352\f\003" offline: false split: false replica_id: 0 } proc_id:
> 81545 }; retry (#8888) after 200805ms delay (Master is coming online...).
> {code}
> The delay doesn't seem correct or respected.
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
