[jira] [Updated] (HBASE-15677) FailedServerException shouldn't clear MetaCache

["Andrew Purtell (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HBASE-15677?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Andrew Purtell updated HBASE-15677:
-----------------------------------
    Fix Version/s:     (was: 1.5.0)
                   1.5.1

> FailedServerException shouldn't clear MetaCache
> -----------------------------------------------
>
>                 Key: HBASE-15677
>                 URL: https://issues.apache.org/jira/browse/HBASE-15677
>             Project: HBase
>          Issue Type: Sub-task
>          Components: Client
>    Affects Versions: 1.3.0
>            Reporter: Mikhail Antonov
>            Assignee: Mikhail Antonov
>            Priority: Major
>             Fix For: 3.0.0, 1.5.1
>
>
> Right now FailedServerException clears meta cache. Seems like it's unnecessary (if we hit that, someone has already gotten some network/remote error in the first place and invalidated located cache for us), and seems it could lead to unnecessary drops, as FailedServers cache has default TTL of 2 seconds, so we can encounter situation like this:
>  - thread T1 hit network error and cleared the cache, put server in failed server list
>  - thread T2 tries to get it's request in and gets FailedServerException
>  - thread T1 does meta scan to populate the cache
>  - thread T2 clears the cache after it's got FSE.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)