Re: Developing cross-component patches post-split

2009-07-01 Thread Raghu Angadi
-1 for committing the jar. Most of the various options proposed sound certainly better. Can build.xml be updated such that Ivy fetches recent (nightly) build? HDFS could have a build target that builds common jar from a specified source location for common. Raghu. Todd Lipcon wrote: On Wed

Re: different modes of inter process communication

2009-09-23 Thread Raghu Angadi
A heartBeat is also an RPC. When you pause Namenode for 30 sec the datanode's heartbeat thread just waits for 30 sec for its heartbeat RPC to return. Note that when you pause Namenode, the RPCs to it don't fail immediately. During this wait, DNs can perform other transactions like serving dat

Re: different modes of inter process communication

2009-09-25 Thread Raghu Angadi
Steve Loughran wrote: Raghu Angadi wrote: A heartBeat is also an RPC. When you pause Namenode for 30 sec the datanode's heartbeat thread just waits for 30 sec for its heartbeat RPC to return. Note that when you pause Namenode, the RPCs to it don't fail immediately. During this wai

Re: Problem reading HDFS block size > 1GB

2009-10-01 Thread Raghu Angadi
Vinay, This issue came up before ( http://www.mail-archive.com/core-...@hadoop.apache.org/msg35620.html ) . I think we should fix this soon. Dhruba filed a jira ( https://issues.apache.org/jira/browse/HDFS-96 ) . Not all errors reported here fixed by the patch attached there. Could we discuss thi

Re: [VOTE] port HADOOP-6218 (Split TFile by Record Sequence Number) to hadoop 0.20/0.21

2009-10-12 Thread Raghu Angadi
+1. risk is low since it does not involve any change to on-disk format. On Mon, Oct 12, 2009 at 4:37 PM, Owen O'Malley wrote: > > On Oct 12, 2009, at 3:55 PM, Hong Tang wrote: > > I suggest that we port the patch back to hadoop 0.20/0.21 now. >> > > This seems like a low risk change, which onl

Review Request : MAPREDUCE-2149 : an improvement to DistCp setup.

2010-11-09 Thread Raghu Angadi
https://issues.apache.org/jira/browse/HADOOP-7023 : Distcp : setup with update is too slow when latency is high This improves 'distcp -update' across geographically separated clusters. All the test pass and patch is tested in cluster. Thanks, Raghu.