hadoop-hdfs-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Fernando O." <fot...@gmail.com>
Subject Re: "hdfs.DFSClient: Could not obtain block"
Date Thu, 13 Oct 2011 14:00:05 GMT
update
checked the logs, and it looks like I shut it down.

2011-10-13 00:51:25,133 WARN
org.apache.hadoop.hdfs.server.namenode.FSNamesystem:
PendingReplicationMonitor timed out block blk_-442888
4440615767535_1052
2011-10-13 00:51:29,249 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* ask
192.168.3.101:50010 to replicate blk_-4428884440615767535_1
052 to datanode(s) 192.168.3.112:50010
2011-10-13 00:51:32,250 INFO org.apache.hadoop.hdfs.StateChange: BLOCK* ask
192.168.3.101:50010 to replicate blk_-4428884440615767535_1
052 to datanode(s) 192.168.3.112:50010
2011-10-13 00:59:07,031 INFO org.apache.hadoop.hdfs.StateChange: DIR*
NameSystem.completeFile: file
/test/testDir2/ferLaptop/testFile2--2011-10-13_00000 is closed by
DFSClient_-268293342
2011-10-13 00:59:08,306 INFO
org.apache.hadoop.hdfs.server.namenode.FSNamesystem: Number of transactions:
34 Total time for transaction
s(ms): 46Number of transactions batched in Syncs: 4 Number of syncs: 0
SyncTimes(ms): 0
2011-10-13 01:00:55,626 INFO
org.apache.hadoop.hdfs.server.namenode.NameNode: SHUTDOWN_MSG:
/************************************************************
SHUTDOWN_MSG: Shutting down NameNode at ferLaptop/127.0.1.1
************************************************************/


On Thu, Oct 13, 2011 at 10:43 AM, Fernando O. <fotero@gmail.com> wrote:

> Hi guys!
>
>     So I'm starting with hadoop. I set up a distributed cluster of 2 nodes
> using a VM. Run a couple of MapReduces, everything went great yesterday.
> When I went to bed I turned off my laptop without excecuting the stop
> scripts and maybe, I don't remember exactly, there were some task writing a
> file (testFile2-2011-10-12_00000 in the log).
>
> When I turn on my machine today I got 1 corrupt block.
>
> I'm pretty sure that if I was not writing a file there's something really
> bad in hdfs. So I assume I was doing it:
>
> It might make sence since I did not stop the server in the correct way and
> I was shutting down the master server (and slave).  On the other hand I
> would expect hadoop not to corrupt my file (the file already existed, I was
> adding data to it) and either to keep the partial state or the previous
> state for it.
>
> So now I'm not sure if I missed something in hadoop's architecture (most
> likely since I didn't read the whoooooole document ), if this is a bug, or
> I'm not seeing the issue (hopefully related with not reading the whole doc )
>
> Thanks.
>
> ps: here's the fsck
>
> $ bin/hdfs fsck /test
> 11/10/13 10:24:22 INFO security.Groups: Group mapping
> impl=org.apache.hadoop.security.ShellBasedUnixGroupsMapping;
> cacheTimeout=300000
> 11/10/13 10:24:22 WARN conf.Configuration: mapred.used.genericoptionsparser
> is deprecated. Instead, use mapreduce.client.genericoptionsparser.used
> .
> /test/testDir1/ferLaptop/testFile1-2011-10-12_00000:  Under replicated
> blk_3656455300577092382_1046. Target Replicas is 3 but found 2 replica(s).
> ...
> /test/testDir2/ferLaptop/testFile2-2011-10-12_00000: CORRUPT block
> blk_-4428884440615767535
>
> /test/testDir2/ferLaptop/testFile2-2011-10-12_00000:  Under replicated
> blk_-4428884440615767535_1052. Target Replicas is 3 but found 1 replica(s).
> .Status: CORRUPT
>  Total size: 26305880 B
>  Total dirs: 5
>  Total files: 5
>  Total blocks (validated): 4 (avg. block size 6576470 B)
>   ********************************
>   CORRUPT FILES: 1
>   CORRUPT BLOCKS: 1
>   ********************************
>  Minimally replicated blocks: 4 (100.0 %)
>  Over-replicated blocks: 0 (0.0 %)
>  Under-replicated blocks: 2 (50.0 %)
>  Mis-replicated blocks: 0 (0.0 %)
>  Default replication factor: 2
>  Average block replication: 1.75
>  Corrupt blocks: 1
>  Missing replicas: 3 (42.857143 %)
>  Number of data-nodes: 2
>  Number of racks: 1
>

Mime
View raw message