[ https://issues.apache.org/jira/browse/HBASE-25743?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17316530#comment-17316530 ]
Bharath Vissapragada commented on HBASE-25743: ---------------------------------------------- > The PR changes OPERATIONTIMEOUT to do retry? Is this a change in behavior? I just collapsed the case statements for CONNECTIONLOSS/OPERATIONTIMEOUT into a single one, no change in functionality? Did I miss something? > Retry REQUESTTIMEOUT KeeperExceptions from ZK > --------------------------------------------- > > Key: HBASE-25743 > URL: https://issues.apache.org/jira/browse/HBASE-25743 > Project: HBase > Issue Type: Bug > Components: Zookeeper > Affects Versions: 3.0.0-alpha-1, 1.7.0, 2.5.0, 2.4.3, 2.3.6 > Reporter: Bharath Vissapragada > Assignee: Bharath Vissapragada > Priority: Major > > Starting ZOOKEEPER-2251, client requests exceeding a timeout can throw a > KeeperException with REQUESTTIMEOUT opcode set. RecoverableZookeeper doesn't > transparently retry in such case. This was causing RS aborts when there is a > flaky ZK quorum member serving slow requests (especially in cases like > rolling upgrades and such). -- This message was sent by Atlassian Jira (v8.3.4#803005)