liubingxing opened a new pull request #4088: URL: https://github.com/apache/hadoop/pull/4088
JIRA: [HDFS-16514](https://issues.apache.org/jira/browse/HDFS-16514) Recently, we used the [Standby Read] feature in our test cluster, and deployed 4 namenode as follow: node1 -> active nn node2 -> standby nn node3 -> observer nn node3 -> observer nn If we set ’dfs.client.failover.random.order=true‘, the client may failover twice and wait a long time to send msync to active namenode. ![image](https://user-images.githubusercontent.com/2844826/159257471-4398ae11-fad3-4aee-8f56-1b89bef2f611.png) I think we can reduce the sleep time of the first several failover based on the number of namenode. For example, if 4 namenode are configured, the sleep time of first three failover operations is set to zero. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org