[
https://issues.apache.org/jira/browse/HBASE-22009?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16796293#comment-16796293
]
Xiang Li commented on HBASE-22009:
----------------------------------
Hi [~xucang], I found this patch seems not workable in branch-1. It makes
TestRSGroupsBasics#testClearDeadServers() failed, not at the verification
points but at afterMethold(), with the following error:
{code:java}
[ERROR] Tests run: 1, Failures: 0, Errors: 1, Skipped: 0, Time elapsed: 15.97 s
<<< FAILURE! - in org.apache.hadoop.hbase.rsgroup.TestRSGroupsBasics
[ERROR]
testClearDeadServers(org.apache.hadoop.hbase.rsgroup.TestRSGroupsBasics) Time
elapsed: 3.59 s <<< ERROR!
org.apache.hadoop.hbase.constraint.ConstraintException:
org.apache.hadoop.hbase.constraint.ConstraintException: Target group is the
same as source group: default
at
org.apache.hadoop.hbase.rsgroup.RSGroupAdminServer.moveServers(RSGroupAdminServer.java:169)
at
org.apache.hadoop.hbase.rsgroup.RSGroupAdminEndpoint.moveServers(RSGroupAdminEndpoint.java:216)
at
org.apache.hadoop.hbase.protobuf.generated.RSGroupAdminProtos$RSGroupAdminService.callMethod(RSGroupAdminProtos.java:13870)
at
org.apache.hadoop.hbase.master.MasterRpcServices.execMasterService(MasterRpcServices.java:711)
at
org.apache.hadoop.hbase.protobuf.generated.MasterProtos$MasterService$2.callBlockingMethod(MasterProtos.java:63436)
at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2380)
at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:124)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:297)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:277)
at
org.apache.hadoop.hbase.rsgroup.TestRSGroupsBasics.afterMethod(TestRSGroupsBasics.java:77)
Caused by: org.apache.hadoop.hbase.ipc.RemoteWithExtrasException:
org.apache.hadoop.hbase.constraint.ConstraintException: Target group is the
same as source group: default
at
org.apache.hadoop.hbase.rsgroup.RSGroupAdminServer.moveServers(RSGroupAdminServer.java:169)
at
org.apache.hadoop.hbase.rsgroup.RSGroupAdminEndpoint.moveServers(RSGroupAdminEndpoint.java:216)
at
org.apache.hadoop.hbase.protobuf.generated.RSGroupAdminProtos$RSGroupAdminService.callMethod(RSGroupAdminProtos.java:13870)
at
org.apache.hadoop.hbase.master.MasterRpcServices.execMasterService(MasterRpcServices.java:711)
at
org.apache.hadoop.hbase.protobuf.generated.MasterProtos$MasterService$2.callBlockingMethod(MasterProtos.java:63436)
at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2380)
at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:124)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:297)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:277)
{code}
I do not figure out the reason yet...
Shall we revert this change from branch-1 as well as 1.5.1 ? Sorry for the
inconvenience it brings
> Improve RSGroupInfoManagerImpl#getDefaultServers()
> --------------------------------------------------
>
> Key: HBASE-22009
> URL: https://issues.apache.org/jira/browse/HBASE-22009
> Project: HBase
> Issue Type: Improvement
> Components: rsgroup
> Reporter: Xiang Li
> Assignee: Xiang Li
> Priority: Minor
> Fix For: 3.0.0, 2.2.0, 1.5.1, 2.2.1
>
> Attachments: HBASE-22009.master.000.patch,
> call_stack_getDefaultServers.png
>
>
> {code:title=RSGroupInfoManagerImpl.java|borderStyle=solid}
> private SortedSet<Address> getDefaultServers() throws IOException {
> SortedSet<Address> defaultServers = Sets.newTreeSet();
> for (ServerName serverName : getOnlineRS()) {
> Address server = Address.fromParts(serverName.getHostname(),
> serverName.getPort());
> boolean found = false;
> for (RSGroupInfo rsgi : listRSGroups()) {
> if (!RSGroupInfo.DEFAULT_GROUP.equals(rsgi.getName()) &&
> rsgi.containsServer(server)) {
> found = true;
> break;
> }
> }
> if (!found) {
> defaultServers.add(server);
> }
> }
> return defaultServers;
> }
> {code}
> That is a logic of 2 nest loops. And for each server, listRSGroups()
> allocates a new LinkedList and calls Map#values(), both of which are very
> heavy operations.
> Maybe the inner loop could be moved out, that is
> # Build a list of servers of other groups than default group
> # Iterate each online servers and check if it is in the list above. If it is
> not, then it belongs to default group.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
