[jira] Updated: (HADOOP-2992) Sequential distributed upgrades.

["Konstantin Shvachko (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HADOOP-2992?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Konstantin Shvachko updated HADOOP-2992:
----------------------------------------

    Attachment: DistrUpgrade.patch

This patch fixes the following problems.
- Ordering of the registered distributed upgrade objects in UpgradeObjectCollection is corrected.
- UpgradeManagerDatanode starts a new upgrade when requested by the name-node.
- I improved the upgrade state reporting.
- TestDistributedUpgrade is testing 3 sequential upgrades now from a pre-upgrade data (LV=-3). 
It uses TestDFSUpgradeFromImage in order to unpack the storage.
- I separate the unpacking method into TestDFSUpgradeFromImage.unpackStorage()
- I also turned off the block scanner for both of these tests, because the name-node is in safe mode most of the time,
and the scanner is just receiving safe-mode exceptions, which only slows down the tests.

> Sequential distributed upgrades.
> --------------------------------
>
>                 Key: HADOOP-2992
>                 URL: https://issues.apache.org/jira/browse/HADOOP-2992
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: dfs, test
>    Affects Versions: 0.14.0
>            Reporter: Konstantin Shvachko
>            Assignee: Konstantin Shvachko
>             Fix For: 0.17.0
>
>         Attachments: DistrUpgrade.patch
>
>
> The distributed upgrades were tested in the case of one distributed upgrade only.
> It turned out to be that current implementation of the framework does not support multiple upgrades.
> This feature is needed for the appends. See HADOOP-2656.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.