You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-user@hadoop.apache.org by Jeff Eastman <je...@collab.net> on 2008/01/15 22:10:54 UTC

Fsck?

Still a noob, so perhaps there is a page somewhere with this
information, but:

 

I had a couple of machines in my cloud go down - not at the same time I
don't think - and they showed up as dead nodes. I bounced the cloud and
ran fsck, which showed a number of 1- and a couple 2- missing blocks for
a number of files. It also said my dfs is corrupted (I assume by the
missing blocks). I've had replication to 3 (12 nodes), and none of the
missing block counts is >2 so I wonder "Are my files hosed?" If they are
it is not a catastrophe, since I have the originals, but I've looked for
fsck documentation and not found an answer.

 

I then tried fsck / -move, which is advertised to move corrupted files
to /lost+found. It printed out a number of file entries and varying ...
afterwards, but there is no /lost+found directory to be found. When I
run fsck / -files again I get the same listing of many good files and
some that are missing 1-2 blocks.

 

I thought I would ask for help before I fsck / -delete to see what will
happen.

Jeff

 

 


RE: Fsck?

Posted by Jeff Eastman <je...@collab.net>.
"Use the code, Jeff"

1) Missing blocks are reported only when all replicas are missing
2) The files are history
3) The dfs won't actually do anything in safe mode
4) Try creating /lost+found first

Jeff

-----Original Message-----
From: Jeff Eastman [mailto:jeastman@collab.net] 
Sent: Tuesday, January 15, 2008 1:11 PM
To: hadoop-user@lucene.apache.org
Subject: Fsck?

Still a noob, so perhaps there is a page somewhere with this
information, but:

 

I had a couple of machines in my cloud go down - not at the same time I
don't think - and they showed up as dead nodes. I bounced the cloud and
ran fsck, which showed a number of 1- and a couple 2- missing blocks for
a number of files. It also said my dfs is corrupted (I assume by the
missing blocks). I've had replication to 3 (12 nodes), and none of the
missing block counts is >2 so I wonder "Are my files hosed?" If they are
it is not a catastrophe, since I have the originals, but I've looked for
fsck documentation and not found an answer.

 

I then tried fsck / -move, which is advertised to move corrupted files
to /lost+found. It printed out a number of file entries and varying ...
afterwards, but there is no /lost+found directory to be found. When I
run fsck / -files again I get the same listing of many good files and
some that are missing 1-2 blocks.

 

I thought I would ask for help before I fsck / -delete to see what will
happen.

Jeff