hadoop-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bogdan M. Maryniuk" <bogdan.maryn...@gmail.com>
Subject Re: Mailing lists are broken?
Date Thu, 02 Jul 2009 08:11:23 GMT
On Thu, Jul 2, 2009 at 3:23 PM, Philip Zeyliger<philip@cloudera.com> wrote:
> That page is currently out of date.  I believe the correct lists are
> {common,hdfs,mapred}-{user,dev,issues}@hadoop.apache.org
> Yes, that's 9 lists.
> There's also general@, and lists for the subprojects (Avro, Hive, Pig,
> HBase, Zookeeper, ...)

Then in this case there must be two updates...:
1. Anyone please update mailing list page to let stop newcomers
cross-post stuff around.
2. Wiki: I can confirm 0.20 works fine on OpenSolaris zones in a true
cluster. :)

...and one question from newbie (me):
By default Hadoop stores stuff in /tmp/.... directory that is "a base
for other temporary directories.", however among that "other
temporary" stuff I see my actual fsimage and VERSION file. Are they
temporary? See, usually /tmp is just 1GB space on a separated
partition and is volatile (e.g. system or node reboot simply wipes it
out). Does that mean that Hadoop moves actual data blocks elsewhere or
actually stores all the stuff actually in /tmp directory?

For some reason, if I set like:


...where I have lots of space per a node, somewhat Hadoop says
capacity 0Kb. :-) When I put anything, it crashes with "File
10mb-random.dat could only be replicated to 0 nodes, instead of 1".

09/07/02 16:33:17 WARN hdfs.DFSClient: Error Recovery for block null
bad datanode[0] nodes == null
09/07/02 16:33:17 WARN hdfs.DFSClient: Could not get block locations.
Source file "/10mb-random.dat" - Aborting...

After I configure it back to "/tmp/anywhere-I-want", everything OK.
Why it is like this? And how to make sure my data is not in a
temporary place?

Kind regards, BM

Things, that are stupid at the beginning, rarely ends up wisely.

View raw message