hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yoram Arnon (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-442) slaves file should include an 'exclude' section, to prevent "bad" datanodes and tasktrackers from disrupting a cluster
Date Thu, 01 Feb 2007 03:04:05 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-442?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12469320
] 

Yoram Arnon commented on HADOOP-442:
------------------------------------

so perhaps a simpler approach, more in line with the original request:

* single slaves file
* no new config variables, since the slaves file is configurable anyway via the hadoop script(s)
and env vars
* lines in the slaves file that start with the word 'exclude' imply that the nodes listed
on that line are to be excluded

pros of this approach:
* completely backwards compatible
* very simple for the administrator (and implementor) - no config, no confusion, single slaves
file to edit

the intent was to simplify the administrator's life. I'd list *all* the cluster in the slaves
file, then list a handful of nodes under an 'exclude' section. The exclude statement is an
override on whatever is listed elsewhere in the file.
The behavior is:
* datanode/tasktracker on nodes that appear in the slaves file and are not excluded are launched
on startup
* any node may connect to the master, unless it is excluded (for backwards compatibility.
I'd actually prefer to only allow included nodes to connect but that will wait, perhaps indefinitely)

nodes are excluded for a reason, basically because they're dead and
* their presence clutters up the startup/shutdown script output and error logs
* if they are included in the cluster they may degrade its behavior or performance so they
should be actively ignored
- I wouldn't worry about the replicas of data they contain - they're long dead and re-replicated.
If they're alive, I won't exclude them.

I love the admin command to reread the file.


> slaves file should include an 'exclude' section, to prevent "bad" datanodes and tasktrackers
from disrupting  a cluster
> -----------------------------------------------------------------------------------------------------------------------
>
>                 Key: HADOOP-442
>                 URL: https://issues.apache.org/jira/browse/HADOOP-442
>             Project: Hadoop
>          Issue Type: Bug
>          Components: conf
>            Reporter: Yoram Arnon
>         Assigned To: Wendy Chien
>
> I recently had a few nodes go bad, such that they were inaccessible to ssh, but were
still running their java processes.
> tasks that executed on them were failing, causing jobs to fail.
> I couldn't stop the java processes, because of the ssh issue, so I was helpless until
I could actually power down these nodes.
> restarting the cluster doesn't help, even when removing the bad nodes from the slaves
file - they just reconnect and are accepted.
> while we plan to avoid tasks from launching on the same nodes over and over, what I'd
like is to be able to prevent rogue processes from connecting to the masters.
> Ideally, the slaves file will contain an 'exclude' section, which will list nodes that
shouldn't be accessed, and should be ignored if they try to connect. That would also help
in configuring the slaves file for a large cluster - I'd list the full range of machines in
the cluster, then list the ones that are down in the 'exclude' section

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


Mime
View raw message