Re: Rolling a Hadoop 0.20.2

2010-01-30 Thread Stack
On Fri, Jan 29, 2010 at 11:26 AM, Hairong Kuang wrote: > TestFiDataTransferProtocol2 tests HDFS-101. I reversed the change made by > HDFS-101 from my local HDFS trunk and ran the test. The test failed. It > seems to me that we still need HDFS-101 in 0.20. > It would seem so. Let me know if you n

Re: Rolling a Hadoop 0.20.2

2010-01-29 Thread Hairong Kuang
TestFiDataTransferProtocol2 tests HDFS-101. I reversed the change made by HDFS-101 from my local HDFS trunk and ran the test. The test failed. It seems to me that we still need HDFS-101 in 0.20. Hairong On 1/29/10 10:10 AM, "Hairong Kuang" wrote: > Stack, > > Could you please check if hbase st

Re: Rolling a Hadoop 0.20.2

2010-01-29 Thread Hairong Kuang
Stack, Could you please check if hbase still needs HDFS-101? I am not sure if HDFS-793 itself solves the problem that HDFS-101 intends to solve. Hairong On 1/29/10 8:55 AM, "Owen O'Malley" wrote: > > On Jan 26, 2010, at 7:09 PM, Todd Lipcon wrote: > >> Owen: is there a particular rush for 0.

Re: Rolling a Hadoop 0.20.2

2010-01-29 Thread Owen O'Malley
On Jan 26, 2010, at 7:09 PM, Todd Lipcon wrote: Owen: is there a particular rush for 0.20.2 or can we hold out for these patches? Pig needs one of the fixes in 0.20.2, so they would like it to be sooner rather than later. It sounds like getting HDFS-127 committed to 0.20.2, so let's see

Re: Rolling a Hadoop 0.20.2

2010-01-27 Thread Todd Lipcon
I don't think 101 is a real blocker - I haven't been able to reliably produce it. HDFS-793 fixes most of the issues I've seen in practice, and we now have a new patch in branch-20 which should fix that. 127 would be nice - there's a patch on that JIRA waiting for review as below. -Todd On Wed, J

Re: Rolling a Hadoop 0.20.2

2010-01-27 Thread Stack
If the new release included (fixedup) hdfs-127 and hdfs-101, the hbase crew would be big fans of an hadoop 0.20.2. St.Ack On Tue, Jan 26, 2010 at 7:09 PM, Todd Lipcon wrote: > HDFS-127 was originally committed, then rolled back from 0.20 as it caused > test timeouts due to infinite loop. I put a

Re: Rolling a Hadoop 0.20.2

2010-01-26 Thread Todd Lipcon
HDFS-127 was originally committed, then rolled back from 0.20 as it caused test timeouts due to infinite loop. I put a new patch up there but it has not been reviewed, and thus not committed. Owen: is there a particular rush for 0.20.2 or can we hold out for these patches? -Todd 2010/1/26 Kay Ka

Re: Rolling a Hadoop 0.20.2

2010-01-26 Thread Kay Kay
Is HDFS-127 going to be part of it ? (seems to have been committed as per the jira). On 1/26/10 6:53 PM, Konstantin Boudnik wrote: +1 On Tue, Jan 26, 2010 at 09:56AM, Owen O'Malley wrote: I'm planning on rolling a Hadoop 0.20.2 today. Are there any blockers that can't wait? -- Owen

Re: Rolling a Hadoop 0.20.2

2010-01-26 Thread Konstantin Boudnik
+1 On Tue, Jan 26, 2010 at 09:56AM, Owen O'Malley wrote: > I'm planning on rolling a Hadoop 0.20.2 today. Are there any blockers > that can't wait? > > -- Owen

Re: Rolling a Hadoop 0.20.2

2010-01-26 Thread Todd Lipcon
In my opinion, I think we should apply the new patch for HDFS-127. Also, there hasn't been a new patch made yet for HDFS-101. Not sure if these should be classified as blockers, since they're not regressions, but I do think they need to go into a branch-20 release at some point. HDFS-127 is critic