Perhaps of note, a mail in the archive with subject "HDFS unbalance issue. (HBase over HDFS)" fingered HADOOP-5124 -- committed to hadoop trunk -- as a help with block imbalance. St.Ack
On Wed, May 13, 2009 at 3:44 PM, Alexandra Alecu <[email protected]>wrote: > > Thank you very much for your answers. > > > Andrew Purtell-2 wrote: > > > > I notice the same behavior, both with uneven block placement, and errors > > running the balancer. DFS will wait to do the deletions until some time > > when the cluster is lightly loaded. I believe there is a config setting > > which can alter that behavior, but I do not recall it offhand. Regarding > > balancer errors, when I see this I restart and that clears the problem. I > > wonder if it is some bug, but have not had time to dig in. > > > > Andy, I checked the hadoop available config settings and I was wondering if > the right setting I need to change for this is fs.trash.interval. I was > thinking to set it to something like 10 min for the import only and then > return to the default. What are your recommendations for this ? > > Thanks, > Alexandra. > > > -- > View this message in context: > http://www.nabble.com/Hbase-0.19.2---Large-import-results-in-heavily-unbalanced-hadoop-DFS-tp23526652p23531320.html > Sent from the HBase User mailing list archive at Nabble.com. > >
