[ https://issues.apache.org/jira/browse/CASSANDRA-8595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14314711#comment-14314711 ]
sankalp kohli commented on CASSANDRA-8595: ------------------------------------------ During reads, we block on a latch to count the number of responses. You need to do some code changes to figure out which endpoint did not reply. You also need to account for speculative retry. You can start from SP.fetchRows method to see how reads happen. > Emit timeouts per endpoint > -------------------------- > > Key: CASSANDRA-8595 > URL: https://issues.apache.org/jira/browse/CASSANDRA-8595 > Project: Cassandra > Issue Type: Improvement > Reporter: sankalp kohli > Priority: Minor > > We currently emit number of timeouts experienced by a co-ordinator while > doing reads and writes. This does not tell us which replica or endpoint is > responsible for the timeouts. > We can keep a map of endpoint to number of timeouts which could be emitted > via JMX. -- This message was sent by Atlassian JIRA (v6.3.4#6332)