[jira] [Commented] (HDFS-4516) Client crash after block allocation and NN switch before lease recovery for the same file can cause readers to fail forever

2013-11-21 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4516?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13829594#comment-13829594 ] Han Xiao commented on HDFS-4516: The revision is elegant. We are benefit from the modified

[jira] [Commented] (HDFS-4154) BKJM: Two namenodes usng bkjm can race to create the version znode

2013-05-22 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13663967#comment-13663967 ] Han Xiao commented on HDFS-4154: Sorry for the format before. Hi, Uma There has been a tes

[jira] [Commented] (HDFS-4154) BKJM: Two namenodes usng bkjm can race to create the version znode

2013-05-22 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13663962#comment-13663962 ] Han Xiao commented on HDFS-4154: Hi, Uma There has been a testcase for the concurrentFormat

[jira] [Commented] (HDFS-4750) Support NFSv3 interface to HDFS

2013-04-28 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4750?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13643948#comment-13643948 ] Han Xiao commented on HDFS-4750: This feature is intersting. However, I'am sorry for not h

[jira] [Commented] (HDFS-4154) BKJM: Two namenodes usng bkjm can race to create the version znode

2013-04-07 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13625138#comment-13625138 ] Han Xiao commented on HDFS-4154: Ok, i will do it. > BKJM: Two namenodes u

[jira] [Updated] (HDFS-4516) Client crash after block allocation and NN switch before lease recovery for the same file can cause readers to fail forever

2013-03-21 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4516?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4516: --- Attachment: HDFS-4516.txt raw patch > Client crash after block allocation and NN switch befor

[jira] [Commented] (HDFS-4516) Client crash after block allocation and NN switch before lease recovery for the same file can cause readers to fail forever

2013-03-20 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4516?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13607386#comment-13607386 ] Han Xiao commented on HDFS-4516: i perfer to add confirmAddBlock since that make fsync not

[jira] [Commented] (HDFS-4516) Client crash after block allocation and NN switch before lease recovery for the same file can cause readers to fail forever

2013-03-20 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4516?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13607379#comment-13607379 ] Han Xiao commented on HDFS-4516: Yes, i also find this problem. We need another rpc method

[jira] [Commented] (HDFS-4516) Client crash after block allocation and NN switch before lease recovery for the same file can cause readers to fail forever

2013-03-18 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4516?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13605098#comment-13605098 ] Han Xiao commented on HDFS-4516: For second method, we could revise like that: 1. for serve

[jira] [Commented] (HDFS-4516) Client crash after block allocation and NN switch before lease recovery for the same file can cause readers to fail forever

2013-03-18 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4516?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13604963#comment-13604963 ] Han Xiao commented on HDFS-4516: Hi Uma. For the first method. There is also problem that t

[jira] [Commented] (HDFS-4608) A File could not be recovery, when a block is added to it in nn, but not created in dn.

2013-03-18 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4608?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13604964#comment-13604964 ] Han Xiao commented on HDFS-4608: duplated to HDFS-4516 > A File could not

[jira] [Updated] (HDFS-4608) A File could not be recovery, when a block is added to it in nn, but not created in dn.

2013-03-18 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4608?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4608: --- Description: In such steps. A file would not be recoveried forever. 1. create a file 2. write data and flush 3

[jira] [Commented] (HDFS-4608) A File could not be recovery, when a block is added to it in nn, but not created in dn.

2013-03-18 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4608?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13604922#comment-13604922 ] Han Xiao commented on HDFS-4608: if the NN not restarted in such a senario, the problem wou

[jira] [Updated] (HDFS-4608) A File could not be recovery, when a block is added to it in nn, but not created in dn.

2013-03-18 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4608?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4608: --- Description: In such steps. A file would not be recoveried forever. 1. create a file 2. write data and flush 3

[jira] [Created] (HDFS-4608) A File could not be recovery, when a block is added to it in nn, but not created in dn.

2013-03-18 Thread Han Xiao (JIRA)
Han Xiao created HDFS-4608: -- Summary: A File could not be recovery, when a block is added to it in nn, but not created in dn. Key: HDFS-4608 URL: https://issues.apache.org/jira/browse/HDFS-4608 Project: Hado

[jira] [Commented] (HDFS-4154) BKJM: Two namenodes usng bkjm can race to create the version znode

2012-12-31 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13541326#comment-13541326 ] Han Xiao commented on HDFS-4154: sorry for updating the patch until now. :D

[jira] [Updated] (HDFS-4154) BKJM: Two namenodes usng bkjm can race to create the version znode

2012-12-31 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4154?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4154: --- Attachment: HDFS-4154.patch > BKJM: Two namenodes usng bkjm can race to create the version znode > ---

[jira] [Commented] (HDFS-4154) BKJM: Two namenodes usng bkjm can race to create the version znode

2012-12-31 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13541315#comment-13541315 ] Han Xiao commented on HDFS-4154: Ok. > BKJM: Two namenodes usng bkjm can r

[jira] [Updated] (HDFS-4130) BKJM: The reading for editlog at NN starting using bkjm is not efficient

2012-12-10 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Fix Version/s: 2.0.3-alpha Hadoop Flags: Reviewed Status: Patch Available (was: In Progress)

[jira] [Work started] (HDFS-4130) BKJM: The reading for editlog at NN starting using bkjm is not efficient

2012-12-10 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Work on HDFS-4130 started by Han Xiao. > BKJM: The reading for editlog at NN starting using bkjm is not efficient > --

[jira] [Commented] (HDFS-4154) BKJM: Two namenodes usng bkjm can race to create the version znode

2012-11-15 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13498530#comment-13498530 ] Han Xiao commented on HDFS-4154: We have repair it in our internal version. I could upload

[jira] [Commented] (HDFS-3908) In HA mode, when there is a ledger in BK missing, which is generated after the last checkpoint, NN can not restore it.

2012-11-15 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3908?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13498524#comment-13498524 ] Han Xiao commented on HDFS-3908: @Uma You are right for the understanding. Two points: 1. I

[jira] [Updated] (HDFS-4130) BKJM: The reading for editlog at NN starting using bkjm is not efficient

2012-11-15 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Attachment: HDFS-4130-v3.patch > BKJM: The reading for editlog at NN starting using bkjm is not efficient

[jira] [Updated] (HDFS-4130) BKJM: The reading for editlog at NN starting using bkjm is not efficient

2012-11-15 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Attachment: (was: HDFS-4130-v3.patch) > BKJM: The reading for editlog at NN starting using bkjm is no

[jira] [Updated] (HDFS-4130) BKJM: The reading for editlog at NN starting using bkjm is not efficient

2012-11-15 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Attachment: HDFS-4130-v3.patch Reasonable suggestion. Accept it. > BKJM: The reading for edit

[jira] [Updated] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-11-09 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Attachment: HDFS-4130-v2.patch Nice suggestion, thanks for Uma. > The reading for editlog at

[jira] [Commented] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-11-05 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13491147#comment-13491147 ] Han Xiao commented on HDFS-4130: Hi Uma, There are thousands of them. The default configur

[jira] [Updated] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-10-30 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Description: Now, the method of BookKeeperJournalManager.selectInputStreams is written like: while (true)

[jira] [Updated] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-10-30 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Description: Now, the method of BookKeeperJournalManager.selectInputStreams is written like: while (true)

[jira] [Updated] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-10-30 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Attachment: HDFS-4130.patch patch available > The reading for editlog at NN starting using bk

[jira] [Updated] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-10-30 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Status: Open (was: Patch Available) Not find the option to submit the patch? > The reading f

[jira] [Updated] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-10-30 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-4130?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-4130: --- Target Version/s: 2.0.2-alpha Status: Patch Available (was: Open) > The reading for editlog

[jira] [Created] (HDFS-4130) The reading for editlog at NN starting using bkjm is not efficient

2012-10-30 Thread Han Xiao (JIRA)
Han Xiao created HDFS-4130: -- Summary: The reading for editlog at NN starting using bkjm is not efficient Key: HDFS-4130 URL: https://issues.apache.org/jira/browse/HDFS-4130 Project: Hadoop HDFS Is

[jira] [Commented] (HDFS-3908) In HA mode, when there is a ledger in BK missing, which is generated after the last checkpoint, NN can not restore it.

2012-09-12 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3908?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13453897#comment-13453897 ] Han Xiao commented on HDFS-3908: Hi Vinay, We could trust Shared Storage, but it is not to

[jira] [Commented] (HDFS-3908) In HA mode, when there is a ledger in BK missing, which is generated after the last checkpoint, NN can not restore it.

2012-09-10 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3908?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13451891#comment-13451891 ] Han Xiao commented on HDFS-3908: to Uma: It is different, HDFS-3769 come from write local f

[jira] [Commented] (HDFS-3908) In HA mode, when there is a ledger in BK missing, which is generated after the last checkpoint, NN can not restore it.

2012-09-10 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3908?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13451884#comment-13451884 ] Han Xiao commented on HDFS-3908: I think we could change the loading of editlogs a little w

[jira] [Updated] (HDFS-3908) In HA mode, when there is a ledger in BK missing, which is generated after the last checkpoint, NN can not restore it.

2012-09-10 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3908?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-3908: --- Summary: In HA mode, when there is a ledger in BK missing, which is generated after the last checkpoint, NN ca

[jira] [Updated] (HDFS-3908) In HA mode, when there is a ledger in BK, which is generated after the last checkpoint, missing, NN can't recover it.

2012-09-10 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3908?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-3908: --- Description: If not HA, when the num of edits.dir is larger than 1. Missing of one editlog file in a dir will

[jira] [Updated] (HDFS-3908) In HA mode, when there is a ledger in BK, which is generated after the last checkpoint, missing, NN can't recover it.

2012-09-10 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3908?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Han Xiao updated HDFS-3908: --- Description: If not HA, when the num of edits.dir is larger than 1. Missing of one editlog file in a dir will

[jira] [Created] (HDFS-3908) In HA mode, when there is a ledger in BK, which is generated after the last checkpoint, missing, NN can't recover it.

2012-09-10 Thread Han Xiao (JIRA)
Han Xiao created HDFS-3908: -- Summary: In HA mode, when there is a ledger in BK, which is generated after the last checkpoint, missing, NN can't recover it. Key: HDFS-3908 URL: https://issues.apache.org/jira/browse/HDFS-3

[jira] [Commented] (HDFS-3586) Blocks are not getting replicate even DN's are availble.

2012-08-13 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3586?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13433835#comment-13433835 ] Han Xiao commented on HDFS-3586: Uma, you are right. In my comments, there also be a condit

[jira] [Commented] (HDFS-3586) Blocks are not getting replicate even DN's are availble.

2012-08-09 Thread Han Xiao (JIRA)
[ https://issues.apache.org/jira/browse/HDFS-3586?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13431831#comment-13431831 ] Han Xiao commented on HDFS-3586: Hi Uma, why not provide a configuration item. Then user ca