[jira] [Created] (HBASE-26623) Report CallDroppedException in exception metrics
Richard Marscher created HBASE-26623: Summary: Report CallDroppedException in exception metrics Key: HBASE-26623 URL: https://issues.apache.org/jira/browse/HBASE-26623 Project: HBase Issue Type: Improvement Components: IPC/RPC, metrics, rpc Reporter: Richard Marscher Assignee: Richard Marscher `CallDroppedException` can be thrown with `CallRunner.drop()` by queue implementations that decide to drop calls to groom the RPC call backlog. The LifoCoDel queue does this I believe and with Pluggable queue it's possible for 3rd party queue implementations to be using `drop()` for similar reasons (my use case). It would be nice for the server to be tracking these exceptions in metrics since otherwise you might have to do some extra lifting on the client side. I have a patch I will submit soon. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (HBASE-22280) Separate read/write handler for priority request(especially for meta).
[ https://issues.apache.org/jira/browse/HBASE-22280?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17463389#comment-17463389 ] Richard Marscher commented on HBASE-22280: -- I also noticed that this means the meta call queue type now becomes the same as your normal rpc call queue type. This is since instead of passing in a FIFO queue like the fast path executor path does, the `MetaRWQueueRpcExecutor` is a child of `RpcExecutor`, which runs through the logic to load the call queue type based on the `base.ipc.server.callqueue.type` config. > Separate read/write handler for priority request(especially for meta). > -- > > Key: HBASE-22280 > URL: https://issues.apache.org/jira/browse/HBASE-22280 > Project: HBase > Issue Type: New Feature > Components: Scheduler >Reporter: Lijin Bin >Assignee: Lijin Bin >Priority: Major > Fix For: 3.0.0-alpha-1, 2.3.0, 2.2.3 > > Attachments: HBASE-22280.patch > > > Client may give too many read pressure on meta, so blocking master write meta > for region open. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Work started] (HBASE-26576) Allow Pluggable Queue to belong to FastPath or normal Balanced Executor
[ https://issues.apache.org/jira/browse/HBASE-26576?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Work on HBASE-26576 started by Richard Marscher. > Allow Pluggable Queue to belong to FastPath or normal Balanced Executor > --- > > Key: HBASE-26576 > URL: https://issues.apache.org/jira/browse/HBASE-26576 > Project: HBase > Issue Type: New Feature > Components: regionserver, rpc >Reporter: Richard Marscher >Assignee: Richard Marscher >Priority: Minor > > The PluggableQueue implementation added in > [https://github.com/apache/hbase/pull/3522] for > https://issues.apache.org/jira/browse/HBASE-6908 is only wired up with the > BalancedQueueRpcExecutor. Users should be able to decide their queue operates > with the FastPathBalancedQueueRpcExecutor optionally via configuration. > I plan to submit a PR for this update soon. > -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Updated] (HBASE-26576) Allow Pluggable Queue to belong to FastPath or normal Balanced Executor
[ https://issues.apache.org/jira/browse/HBASE-26576?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Richard Marscher updated HBASE-26576: - Description: The PluggableQueue implementation added in [https://github.com/apache/hbase/pull/3522] for https://issues.apache.org/jira/browse/HBASE-6908 is only wired up with the BalancedQueueRpcExecutor. Users should be able to decide their queue operates with the FastPathBalancedQueueRpcExecutor optionally via configuration. I plan to submit a PR for this update soon. was: The PluggableQueue implementation added in[ https://github.com/apache/hbase/pull/3522|https://github.com/apache/hbase/pull/3522)] for https://issues.apache.org/jira/browse/HBASE-6908 is only wired up with the BalancedQueueRpcExecutor. Users should be able to decide their queue operates with the FastPathBalancedQueueRpcExecutor optionally via configuration. I plan to submit a PR for this update soon. > Allow Pluggable Queue to belong to FastPath or normal Balanced Executor > --- > > Key: HBASE-26576 > URL: https://issues.apache.org/jira/browse/HBASE-26576 > Project: HBase > Issue Type: New Feature > Components: regionserver, rpc >Reporter: Richard Marscher >Assignee: Richard Marscher >Priority: Minor > > The PluggableQueue implementation added in > [https://github.com/apache/hbase/pull/3522] for > https://issues.apache.org/jira/browse/HBASE-6908 is only wired up with the > BalancedQueueRpcExecutor. Users should be able to decide their queue operates > with the FastPathBalancedQueueRpcExecutor optionally via configuration. > I plan to submit a PR for this update soon. > -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Created] (HBASE-26576) Allow Pluggable Queue to belong to FastPath or normal Balanced Executor
Richard Marscher created HBASE-26576: Summary: Allow Pluggable Queue to belong to FastPath or normal Balanced Executor Key: HBASE-26576 URL: https://issues.apache.org/jira/browse/HBASE-26576 Project: HBase Issue Type: New Feature Components: regionserver, rpc Reporter: Richard Marscher Assignee: Richard Marscher The PluggableQueue implementation added in[ https://github.com/apache/hbase/pull/3522|https://github.com/apache/hbase/pull/3522)] for https://issues.apache.org/jira/browse/HBASE-6908 is only wired up with the BalancedQueueRpcExecutor. Users should be able to decide their queue operates with the FastPathBalancedQueueRpcExecutor optionally via configuration. I plan to submit a PR for this update soon. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Commented] (HBASE-6908) Pluggable Call BlockingQueue for HBaseServer
[ https://issues.apache.org/jira/browse/HBASE-6908?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17385756#comment-17385756 ] Richard Marscher commented on HBASE-6908: - Looks like it auto linked. I submitted a pull request proposal for a solution: https://github.com/apache/hbase/pull/3522 > Pluggable Call BlockingQueue for HBaseServer > > > Key: HBASE-6908 > URL: https://issues.apache.org/jira/browse/HBASE-6908 > Project: HBase > Issue Type: New Feature > Components: IPC/RPC >Reporter: James Taylor >Priority: Major > > Allow the BlockingQueue implementation class to be specified in the HBase > config to enable different behavior than a FIFO queue. The use case we have > is around fairness and starvation for big scans that are parallelized on the > client. It's easy to fill up the HBase server Call BlockingQueue when > processing a single parallelized scan, leadng other scans to time out. > Instead, doing round robin processesing on a dequeue through a different > BlockingQueue implementation will prevent this from occurring. -- This message was sent by Atlassian Jira (v8.3.4#803005)