hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mikhail Antonov (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-17032) CallQueueTooBigException and CallDroppedException should not be triggering PFFE
Date Sat, 05 Nov 2016 00:56:58 GMT

     [ https://issues.apache.org/jira/browse/HBASE-17032?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Mikhail Antonov updated HBASE-17032:
    Attachment: HBASE-17032.branch-1.3.v2.patch

v2 patch with fixed test

> CallQueueTooBigException and CallDroppedException should not be triggering PFFE
> -------------------------------------------------------------------------------
>                 Key: HBASE-17032
>                 URL: https://issues.apache.org/jira/browse/HBASE-17032
>             Project: HBase
>          Issue Type: Bug
>          Components: Client
>    Affects Versions: 1.3.0
>            Reporter: Mikhail Antonov
>            Assignee: Mikhail Antonov
>             Fix For: 2.0.0, 1.3.0
>         Attachments: HBASE-17032.branch-1.3.v1.patch, HBASE-17032.branch-1.3.v2.patch
> Back in HBASE-15137 we made it so that CQTBE causes preemptive fast fail exception on
the client. 
> It seems those 2 load control mechanists don't exactly align here. Server throws CallQueueTooBigException,
CallDroppedException (from deadline scheduler) when it feels overloaded. Client should accept
that behavior and retry. When servers sheds the load, and client also bails out, the load
shedding  bubbles up too high and high level impact on the client applications seems worse
with PFFE turned on then without.

This message was sent by Atlassian JIRA

View raw message