I have just come across a problem that I can't seem to find much info on in The Google.
I have one slave node (v 0.20.1) that is repeatedly filling up one of its disks with the same log message from the DataBlockScanner that reads:
2011-01-06 04:03:26,743 INFO org.apache.hadoop.hdfs.server.datanode.DataBlockScanner: Verification failed for blk_-*_*. Its ok since it not in datanode dataset anymore.
The same error gets written to the log file hundreds of times per second, filling up the disk pretty quickly.
Is there a polite way to expunge a block from the system other than just deleting it on the filesystem? I checked the code, and it looks like it attempts to delete a block like this under normal circumstances, but for whatever reason, this block isn't going away.