You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "eric baldeschwieler (JIRA)" <ji...@apache.org> on 2007/03/06 01:46:24 UTC
[jira] Created: (HADOOP-1065) DFS client warnings / info's should
be centrally logged by name node
DFS client warnings / info's should be centrally logged by name node
--------------------------------------------------------------------
Key: HADOOP-1065
URL: https://issues.apache.org/jira/browse/HADOOP-1065
Project: Hadoop
Issue Type: Improvement
Reporter: eric baldeschwieler
When a client encounters an error condition that it warns / infos the user about, we should log that centrally, so that we can mine this info later. For example CRC errors or block write retries.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-1065) DFS client warnings / info's should
be centrally logged by name node
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-1065?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-1065:
-------------------------------------
Fix Version/s: 0.15.0
> DFS client warnings / info's should be centrally logged by name node
> --------------------------------------------------------------------
>
> Key: HADOOP-1065
> URL: https://issues.apache.org/jira/browse/HADOOP-1065
> Project: Hadoop
> Issue Type: Improvement
> Components: dfs
> Reporter: eric baldeschwieler
> Fix For: 0.15.0
>
>
> When a client encounters an error condition that it warns / infos the user about, we should log that centrally, so that we can mine this info later. For example CRC errors or block write retries.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-1065) DFS client warnings / info's should
be centrally logged by name node
Posted by "eric baldeschwieler (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-1065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12478870 ]
eric baldeschwieler commented on HADOOP-1065:
---------------------------------------------
Sounds like we should always log a reason with a deletion
> DFS client warnings / info's should be centrally logged by name node
> --------------------------------------------------------------------
>
> Key: HADOOP-1065
> URL: https://issues.apache.org/jira/browse/HADOOP-1065
> Project: Hadoop
> Issue Type: Improvement
> Reporter: eric baldeschwieler
>
> When a client encounters an error condition that it warns / infos the user about, we should log that centrally, so that we can mine this info later. For example CRC errors or block write retries.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-1065) DFS client warnings / info's should
be centrally logged by name node
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-1065?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-1065:
-------------------------------------
Component/s: dfs
> DFS client warnings / info's should be centrally logged by name node
> --------------------------------------------------------------------
>
> Key: HADOOP-1065
> URL: https://issues.apache.org/jira/browse/HADOOP-1065
> Project: Hadoop
> Issue Type: Improvement
> Components: dfs
> Reporter: eric baldeschwieler
>
> When a client encounters an error condition that it warns / infos the user about, we should log that centrally, so that we can mine this info later. For example CRC errors or block write retries.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-1065) DFS client warnings / info's should
be centrally logged by name node
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-1065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12478494 ]
dhruba borthakur commented on HADOOP-1065:
------------------------------------------
The namenode currently exposes a RPC called reportBadBlock() that allows a client to report blocks that have bad crc. The namenode silently deletes those bad blocks. Oops!
In the current implementation, the client does not report failed read/write retry attempts to the namenode. It should.
> DFS client warnings / info's should be centrally logged by name node
> --------------------------------------------------------------------
>
> Key: HADOOP-1065
> URL: https://issues.apache.org/jira/browse/HADOOP-1065
> Project: Hadoop
> Issue Type: Improvement
> Reporter: eric baldeschwieler
>
> When a client encounters an error condition that it warns / infos the user about, we should log that centrally, so that we can mine this info later. For example CRC errors or block write retries.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-1065) DFS client warnings / info's should
be centrally logged by name node
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-1065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12478833 ]
dhruba borthakur commented on HADOOP-1065:
------------------------------------------
The reportBadBlock was introduced as part of the 855. The Namenode deletes the blocks that are reported to be bad. It does log it as part of "deletion". My comment was regarding that fact that maybe we can enhance the logging to explicitly state "blocks being deleted because they are bad". Otherwise, from the Namenode log it might be difficult to distinguish blocks that are deleted because of other reasons (decrease of replication factor, file getting deleted, etc) from the case when blocks get deleted because they are detected to be bad
> DFS client warnings / info's should be centrally logged by name node
> --------------------------------------------------------------------
>
> Key: HADOOP-1065
> URL: https://issues.apache.org/jira/browse/HADOOP-1065
> Project: Hadoop
> Issue Type: Improvement
> Reporter: eric baldeschwieler
>
> When a client encounters an error condition that it warns / infos the user about, we should log that centrally, so that we can mine this info later. For example CRC errors or block write retries.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.