[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15346220#comment-15346220
]
Yu Li commented on HBASE-15900:
---
We were running 0.98.12 with hadoop 2.6.2 for a long time and now running
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15345872#comment-15345872
]
Heng Chen commented on HBASE-15900:
---
After test 2.6.0 with one week, and 2.7.2 with one more week.
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15319877#comment-15319877
]
Heng Chen commented on HBASE-15900:
---
I have upgraded my hbase cluster hadoop version to be 2.6.0, it
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15319797#comment-15319797
]
Mikhail Antonov commented on HBASE-15900:
-
[~chenheng] curious if you had a chance to test your
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15319795#comment-15319795
]
Mikhail Antonov commented on HBASE-15900:
-
For 1.3 I think 2.7.2 should be the default version,
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15319790#comment-15319790
]
Andrew Purtell commented on HBASE-15900:
We also saw long stuck flushes or compactions, and
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15315121#comment-15315121
]
Nick Dimiduk commented on HBASE-15900:
--
bq. any plan to upgrade branch-1.1 hadoop version to 2.6.0
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15309944#comment-15309944
]
Sergey Soldatov commented on HBASE-15900:
-
Well, there are a lot of possibilities to block
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15309926#comment-15309926
]
Heng Chen commented on HBASE-15900:
---
{code}
do you happen to have any steps to reliably (or, well, with
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15309910#comment-15309910
]
Mikhail Antonov commented on HBASE-15900:
-
Hm. I _think_ I've seen stuck memstores on 2.7+
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15309900#comment-15309900
]
Mikhail Antonov commented on HBASE-15900:
-
As in the support / compatibility matrix we say that
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15309393#comment-15309393
]
Heng Chen commented on HBASE-15900:
---
branch-1.3 seems compile with 2.5.x too. [~mantonov]
> RS stuck
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15309386#comment-15309386
]
Heng Chen commented on HBASE-15900:
---
Yeah , it looks like this bug block the longCompaction thread.
I
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15309259#comment-15309259
]
Heng Chen commented on HBASE-15900:
---
Maybe. HDFS version is 2.5.1. Let me try to upgrade my hdfs
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15308776#comment-15308776
]
Sergey Soldatov commented on HBASE-15900:
-
[~chenheng] Which HDFS version you are using? It may
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15307450#comment-15307450
]
Heng Chen commented on HBASE-15900:
---
We could see in 0d32a6bab354e6cc170cd59a2d485797.rs.log, the
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15307438#comment-15307438
]
Heng Chen commented on HBASE-15900:
---
I open DEBUG Level log in RS, and found the problem which i
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306269#comment-15306269
]
Heng Chen commented on HBASE-15900:
---
1.8.0_20
I can find threads which own the read lock, but can't
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306264#comment-15306264
]
Sergey Soldatov commented on HBASE-15900:
-
[~chenheng] Thank you. Quite interesting. A bunch of
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306195#comment-15306195
]
Sergey Soldatov commented on HBASE-15900:
-
[~chenheng] that would be great. Thank you!
> RS
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306194#comment-15306194
]
Heng Chen commented on HBASE-15900:
---
Sorry, [~sergey.soldatov] My RS has been restart, but it always
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306192#comment-15306192
]
Sergey Soldatov commented on HBASE-15900:
-
[~chenheng] yet again, could you use jstack with -l
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306187#comment-15306187
]
Heng Chen commented on HBASE-15900:
---
[~stack]
I have found something
HStore.lock.readLock was hold by
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306119#comment-15306119
]
stack commented on HBASE-15900:
---
I didn't know about the '-l'. It is not on mac os x it seems, only on
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15303921#comment-15303921
]
Sergey Soldatov commented on HBASE-15900:
-
I would highly recommend to try get dump using
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15303771#comment-15303771
]
Heng Chen commented on HBASE-15900:
---
In the log, we can see, before request flush of region
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15303759#comment-15303759
]
Heng Chen commented on HBASE-15900:
---
Yes, something like this. I extract one region log which always
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15303751#comment-15303751
]
Mikhail Antonov commented on HBASE-15900:
-
I've seen that as well, for sure (w/o running MR
[
https://issues.apache.org/jira/browse/HBASE-15900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15303748#comment-15303748
]
Heng Chen commented on HBASE-15900:
---
Check code on HStore.java, the method which may hold write lock
29 matches
Mail list logo
