You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kudu.apache.org by "Mike Percy (JIRA)" <ji...@apache.org> on 2016/02/26 13:08:18 UTC

[jira] [Updated] (KUDU-708) Java client contention on KuduSession lock

     [ https://issues.apache.org/jira/browse/KUDU-708?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Mike Percy updated KUDU-708:
----------------------------
    Parent: KUDU-435

> Java client contention on KuduSession lock
> ------------------------------------------
>
>                 Key: KUDU-708
>                 URL: https://issues.apache.org/jira/browse/KUDU-708
>             Project: Kudu
>          Issue Type: Sub-task
>          Components: client
>    Affects Versions: Private Beta
>            Reporter: Todd Lipcon
>            Assignee: Jean-Daniel Cryans
>            Priority: Critical
>
> The client can't seem to use more than one thread very effectively when doing large writes. I see all of the threads blocked in stacks like:
> "New I/O worker #4" prio=10 tid=0x00007fee58044000 nid=0x4c0d waiting for monitor entry [0x00007fee70641000]
>    java.lang.Thread.State: BLOCKED (on object monitor)
>         at kudu.rpc.KuduSession.tabletInFlightDone(KuduSession.java:593)
>         - waiting to lock <0x000000008b4bedc8> (a kudu.rpc.KuduSession)
>         at kudu.rpc.KuduSession.access$500(KuduSession.java:67)
>         at kudu.rpc.KuduSession$2.call(KuduSession.java:570)
>         at kudu.rpc.KuduSession$2.call(KuduSession.java:567)
>         at com.stumbleupon.async.Deferred.doCall(Deferred.java:1278)
>         at com.stumbleupon.async.Deferred.runCallbacks(Deferred.java:1257)
>         at com.stumbleupon.async.Deferred.callback(Deferred.java:1005)
>         at kudu.rpc.KuduRpc.handleCallback(KuduRpc.java:169)
>         at kudu.rpc.KuduRpc.callback(KuduRpc.java:182)
>         at kudu.rpc.TabletClient.decode(TabletClient.java:413)
>         at kudu.rpc.TabletClient.decode(TabletClient.java:82)
> while another thread is holding that lock while doing CPU-heavy work of request encoding. Can we drop the lock during the request encoding work?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)