Billy Pearson over in hadoop-user reports that a bunch of new DFSClient
INFO-level logging has started showing in hbase logs where previous
there was none.
HADOOP-1707 committed a few days ago added the below new LOG.infos:
$ svn diff -r612699 DFSClient.java |grep LOG.info
+ LOG.info("Allocating new block");
+ LOG.info("Closing old block " + block);
+ LOG.info("Error Recovery for block " + block +
+ LOG.info("Abandoning block " + block);
- LOG.info("Waiting to find target node: " + target);
+ LOG.info("Waiting to find target node: " +
nodes[0].getName());
+ LOG.info("pipeline = " + nodes[i].getName());
+ LOG.info("Connecting to " + nodes[0].getName());
+ LOG.info("Exception in createBlockOutputStream " + ie);
- LOG.info("Could not complete file, retrying...");
+ LOG.info("Could not complete file " + src + "
retrying...");
Was it intentional that above be logged at INFO level? If so, we'll
change hbase logging config. Otherwise, I can make a patch to change
the above to DEBUGs instead.
Thanks,
St.Ack
--- Begin Message ---
I just download trunk and started testing and I get this in hbase log
every block written show these 4 lines this makes the hbase log way to hard
to watch as process happen
2008-01-21 02:35:40,675 INFO org.apache.hadoop.fs.DFSClient: Allocating new
block
2008-01-21 02:35:40,680 INFO org.apache.hadoop.fs.DFSClient: pipeline =
10.0.0.4:50010
2008-01-21 02:35:40,680 INFO org.apache.hadoop.fs.DFSClient: pipeline =
10.0.0.2:50010
2008-01-21 02:35:40,680 INFO org.apache.hadoop.fs.DFSClient: Connecting to
10.0.0.4:50010
should this level of logging be done with INFO or DEBUG?
Billy
--- End Message ---
