[ https://issues.apache.org/jira/browse/HBASE-21102?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16606496#comment-16606496 ]
huaxiang sun commented on HBASE-21102: -------------------------------------- +1, a nit, the following else branch does not seem necessary. I did a test to verify that. {code} + List<ServerName> usedSNs = new ArrayList<ServerName>(servers.size()); do { int i = RANDOM.nextInt(numServers); sn = servers.get(i); + if (!usedSNs.contains(sn)) { + usedSNs.add(sn); + } else { + continue; + } // Huaxiang: the else branch does not seem necessary. I did a test, with continue, it still // goes through the while conditions. } while (cluster.wouldLowerAvailability(regionInfo, sn) && iterations++ < maxIterations); + if (iterations >= maxIterations) { + // We have reached the max. Means the servers that we collected is still lowering the + // availability + for (ServerName unusedServer : servers) { + if (!usedSNs.contains(unusedServer)) { + // check if any other unused server is there for us to use. + // If so use it. Else we have not other go but to go with one of them + if (!cluster.wouldLowerAvailability(regionInfo, unusedServer)) { + sn = unusedServer; + break; + } + } + } + } {code} > ServerCrashProcedure should select target server where no other replicas > exist for the current region > ----------------------------------------------------------------------------------------------------- > > Key: HBASE-21102 > URL: https://issues.apache.org/jira/browse/HBASE-21102 > Project: HBase > Issue Type: Bug > Components: Region Assignment > Affects Versions: 3.0.0 > Reporter: ramkrishna.s.vasudevan > Assignee: ramkrishna.s.vasudevan > Priority: Major > Attachments: HBASE-21102_1.patch, HBASE-21102_2.patch, > HBASE-21102_3.patch, HBASE-21102_initial.patch > > > Currently when a server with region replica crashes, when the target server > is created for the replica region assignment there is no guarentee that a > server is selected where there is no other replica for the current region > getting assigned. It so happens that currently we do an assignment randomly > and later the LB comes and identifies these cases and again does MOVE for > such regions. It will be better if we can identify target servers at least > minimally ensuring that replicas are not colocated. -- This message was sent by Atlassian JIRA (v7.6.3#76005)