[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16195103#comment-16195103
]
Hudson commented on HBASE-14004:
Results for branch HBASE-18467, done in 4 hr 24 min and counting
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16185802#comment-16185802
]
Hudson commented on HBASE-14004:
FAILURE: Integrated in Jenkins build HBase-Trunk_matrix #3798 (See
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16185726#comment-16185726
]
Hudson commented on HBASE-14004:
FAILURE: Integrated in Jenkins build HBase-2.0 #597 (See
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16185257#comment-16185257
]
Duo Zhang commented on HBASE-14004:
---
Will do. Thanks for the reminding sir.
> [Replication]
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16184854#comment-16184854
]
stack commented on HBASE-14004:
---
[~Apache9] Any chance of a release note here please boss? This is a great
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16170982#comment-16170982
]
Duo Zhang commented on HBASE-14004:
---
OK, will check later.
> [Replication] Inconsistency between
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16170374#comment-16170374
]
Ted Yu commented on HBASE-14004:
TestReplicationSmallTests fails consistently on master.
If I switch to
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16169643#comment-16169643
]
Duo Zhang commented on HBASE-14004:
---
The problem for FSHLog is complicated. One thing is that, syncing
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16169589#comment-16169589
]
Sean Busbey commented on HBASE-14004:
-
is this needed in branch-1? the doc mentions that the
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16168299#comment-16168299
]
stack commented on HBASE-14004:
---
This is a great fix.
> [Replication] Inconsistency between Memstore and
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16168034#comment-16168034
]
Hudson commented on HBASE-14004:
FAILURE: Integrated in Jenkins build HBase-2.0 #518 (See
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16167998#comment-16167998
]
Hudson commented on HBASE-14004:
FAILURE: Integrated in Jenkins build HBase-Trunk_matrix #3719 (See
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16167727#comment-16167727
]
Duo Zhang commented on HBASE-14004:
---
OK, all green. Let me commit.
> [Replication] Inconsistency
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16167720#comment-16167720
]
Hadoop QA commented on HBASE-14004:
---
| (/) *{color:green}+1 overall{color}* |
\\
\\
|| Vote ||
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16167437#comment-16167437
]
Hadoop QA commented on HBASE-14004:
---
| (x) *{color:red}-1 overall{color}* |
\\
\\
|| Vote || Subsystem
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16165857#comment-16165857
]
stack commented on HBASE-14004:
---
[~Apache9] Sorry. I reviewed it a few days ago but forgot to publish the
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16165819#comment-16165819
]
Duo Zhang commented on HBASE-14004:
---
Thanks [~ram_krish] for reviewing. Will commit later.
>
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16164323#comment-16164323
]
ramkrishna.s.vasudevan commented on HBASE-14004:
Will take a look at this @duo zhang.
>
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16164280#comment-16164280
]
Duo Zhang commented on HBASE-14004:
---
Ping [~stack] [~anoopsamjohn] [~ram_krish], this feature is
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16162524#comment-16162524
]
Hadoop QA commented on HBASE-14004:
---
| (/) *{color:green}+1 overall{color}* |
\\
\\
|| Vote ||
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16162439#comment-16162439
]
Duo Zhang commented on HBASE-14004:
---
[~stack] [~anoopsamjohn] PTAL. Thanks.
> [Replication]
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16162335#comment-16162335
]
Duo Zhang commented on HBASE-14004:
---
Good, no test failures. Let me fix the javadoc issues and upload
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16161489#comment-16161489
]
Hadoop QA commented on HBASE-14004:
---
| (x) *{color:red}-1 overall{color}* |
\\
\\
|| Vote || Subsystem
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15625326#comment-15625326
]
Duo Zhang commented on HBASE-14004:
---
I think we need to pick this up.
With AsyncFSWAL, it is not safe
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15149552#comment-15149552
]
stack commented on HBASE-14004:
---
But, as noted elsewhere, this fact does not solve this issue (what made it
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15149497#comment-15149497
]
stack commented on HBASE-14004:
---
bq. We could try rejiggering the order in which memstore gets updated,
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050322#comment-15050322
]
Yu Li commented on HBASE-14004:
---
bq. Yu, is it still a problem after we not using hsync?
My concern was
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050409#comment-15050409
]
Yu Li commented on HBASE-14004:
---
Let me confirm, do you mean we never rollback the memstore even if
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050418#comment-15050418
]
Heng Chen commented on HBASE-14004:
---
{quote}
Let me confirm, do you mean we never rollback the memstore
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050467#comment-15050467
]
Phil Yang commented on HBASE-14004:
---
We can not rollback even if buffered-entries rewrite failed, or we
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050308#comment-15050308
]
Phil Yang commented on HBASE-14004:
---
[~chenheng] it seems that HBASE-14949 only handle replaying logs
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050333#comment-15050333
]
Heng Chen commented on HBASE-14004:
---
That is my plan.
> [Replication] Inconsistency between Memstore
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050332#comment-15050332
]
Heng Chen commented on HBASE-14004:
---
That is my plan.
> [Replication] Inconsistency between Memstore
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050376#comment-15050376
]
Heng Chen commented on HBASE-14004:
---
{quote}
2. WAL logger tries to re-write the buffered entries to a
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050438#comment-15050438
]
Heng Chen commented on HBASE-14004:
---
IMO we should handle this situation, if we failed during recovery
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15048552#comment-15048552
]
Phil Yang commented on HBASE-14004:
---
You are right, we should change the logic of replicator.
And I am
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15049048#comment-15049048
]
Phil Yang commented on HBASE-14004:
---
The reason I asked this question is if we don't have to make sure
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15049947#comment-15049947
]
Heng Chen commented on HBASE-14004:
---
{quote}
1, 2, 3, 4, 5---this is normal order
1, 3, 2, 4, 5---the
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15050165#comment-15050165
]
Phil Yang commented on HBASE-14004:
---
I have revised the doc
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15049009#comment-15049009
]
stack commented on HBASE-14004:
---
Not sure I understand the question. Replication reads the WAL in order and
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046584#comment-15046584
]
Heng Chen commented on HBASE-14004:
---
I make a patch for skip duplicate entries when replay, i found
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046656#comment-15046656
]
stack commented on HBASE-14004:
---
Doc is great. I added a few little notes...
> [Replication] Inconsistency
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15047929#comment-15047929
]
Heng Chen commented on HBASE-14004:
---
{quote}
Have to be careful how we do this. We have elaborate
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15047986#comment-15047986
]
Heng Chen commented on HBASE-14004:
---
{quote}
But I think there should be a follow-up work on data
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15047976#comment-15047976
]
Phil Yang commented on HBASE-14004:
---
{quote}
And as our design, we only use hsync to ensure data
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15048171#comment-15048171
]
Heng Chen commented on HBASE-14004:
---
{quote}
And if we don't use hsync, I think we need not change the
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15048137#comment-15048137
]
Phil Yang commented on HBASE-14004:
---
{quote}
And in this issue, i suppose we just ensure correctness
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15048010#comment-15048010
]
stack commented on HBASE-14004:
---
This is the issue where client can say what durability to apply per edit.
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046414#comment-15046414
]
Heng Chen commented on HBASE-14004:
---
Why we need 'where to end' when recover replication? We just
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046475#comment-15046475
]
Yu Li commented on HBASE-14004:
---
You are right Heng, and Duo also explained about this. Thanks. :-)
>
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046416#comment-15046416
]
Heng Chen commented on HBASE-14004:
---
Why we need 'where to end' when recover replication? We just
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046415#comment-15046415
]
Heng Chen commented on HBASE-14004:
---
Why we need 'where to end' when recover replication? We just
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046422#comment-15046422
]
Duo Zhang commented on HBASE-14004:
---
I do not think we can benefit a lot from storing acked length to
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046439#comment-15046439
]
Yu Li commented on HBASE-14004:
---
bq. HBase uses NN to do fencing which means HMaster will only consider an
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044970#comment-15044970
]
Phil Yang commented on HBASE-14004:
---
Furthermore, I find another issue that since HDFS-744 supported
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044563#comment-15044563
]
Heng Chen commented on HBASE-14004:
---
Looks good. Except one thing.
{code}
If we get timeout error
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044643#comment-15044643
]
Phil Yang commented on HBASE-14004:
---
{quote}
It seems that WAL reader is designed for one WAL, we can't
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044685#comment-15044685
]
Phil Yang commented on HBASE-14004:
---
{quote}
Guess this may be the original design since there is TODO
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044576#comment-15044576
]
Phil Yang commented on HBASE-14004:
---
{quote}
Some entries in new WAL may have smaller seqid than
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044616#comment-15044616
]
He Liangliang commented on HBASE-14004:
---
Sorry for the late reply. What about this approach:
1.
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044666#comment-15044666
]
Heng Chen commented on HBASE-14004:
---
{quote}
You are right, the logic that ignores logs which are
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044603#comment-15044603
]
Heng Chen commented on HBASE-14004:
---
It seems that WAL reader is designed for one WAL, we can't
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044702#comment-15044702
]
He Liangliang commented on HBASE-14004:
---
We don't need to guarantee "avoid replaying or replicating
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044835#comment-15044835
]
Phil Yang commented on HBASE-14004:
---
{quote}
In another word, the client needn't discriminate between
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046226#comment-15046226
]
Yu Li commented on HBASE-14004:
---
bq. My idea may be more fast because it needn't send request to zk
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046286#comment-15046286
]
Phil Yang commented on HBASE-14004:
---
{quote}
I guess we need to persist the acked length to somewhere
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046292#comment-15046292
]
Heng Chen commented on HBASE-14004:
---
{quote}
I guess we need to persist the acked length to somewhere
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046371#comment-15046371
]
Yu Li commented on HBASE-14004:
---
It seems to me that replication queue only records where to *start*
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046385#comment-15046385
]
Yu Li commented on HBASE-14004:
---
Assume this situation:
1. RS received an ack from DFSClient, and record it
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046395#comment-15046395
]
Duo Zhang commented on HBASE-14004:
---
{quote}
If e don't persist this acked length, during failover we
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046398#comment-15046398
]
Duo Zhang commented on HBASE-14004:
---
{quote}
3. Replication queue recovery starts on another RS, where
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046401#comment-15046401
]
Yu Li commented on HBASE-14004:
---
bq. then the file will finally be closed by NN
I don't think *finally be
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15046405#comment-15046405
]
Duo Zhang commented on HBASE-14004:
---
{quote}
I don't think finally be closed is enough, since the
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044318#comment-15044318
]
Heng Chen commented on HBASE-14004:
---
{quote}
If only RS crashed, DNs not, there is no data loss because
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044319#comment-15044319
]
Duo Zhang commented on HBASE-14004:
---
{quote}
Is it means some entries may be already in place (you
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044321#comment-15044321
]
Duo Zhang commented on HBASE-14004:
---
{quote}
If one hlush failed, we close old WAL and write entries
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044308#comment-15044308
]
Heng Chen commented on HBASE-14004:
---
{quote}
To be clear, this means we will hold the WAL.sync request
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044315#comment-15044315
]
Phil Yang commented on HBASE-14004:
---
{quote}
If RS crashed at this time, what will happen?
{quote}
If
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044327#comment-15044327
]
Heng Chen commented on HBASE-14004:
---
{quote}
No. You can close old WAL using any length that larger
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044341#comment-15044341
]
Heng Chen commented on HBASE-14004:
---
{quote}
Now HBase only use hflush, if we add a hsync logic, will
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044380#comment-15044380
]
Phil Yang commented on HBASE-14004:
---
Sure, I'm going to draft a document for summarizing our ideas.
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044368#comment-15044368
]
Heng Chen commented on HBASE-14004:
---
{quote}
btw, could see quite some discussion here and maybe a doc
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044333#comment-15044333
]
Phil Yang commented on HBASE-14004:
---
I have a new concern about the configuration. Now HBase only use
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15044355#comment-15044355
]
Yu Li commented on HBASE-14004:
---
bq. if we add a hsync logic, will there be degradation of performance
IMHO
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15043721#comment-15043721
]
Yu Li commented on HBASE-14004:
---
Nice discussion [~yangzhe1991] and [~Apache9].
FWIW, two questions about
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15043246#comment-15043246
]
Duo Zhang commented on HBASE-14004:
---
Sounds great, an incremental unique id of WAL entry is better
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042765#comment-15042765
]
Phil Yang commented on HBASE-14004:
---
Is it required to use the size of serialized binary data? I don't
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042744#comment-15042744
]
Duo Zhang commented on HBASE-14004:
---
{quote}
Fixing HBase writing path that we should retry logging WAL
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15043657#comment-15043657
]
stack commented on HBASE-14004:
---
bq. ReplicationSource should only read WAL that is hsynced to prevent
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15043664#comment-15043664
]
Duo Zhang commented on HBASE-14004:
---
{quote}
This will require big change in how replication works but
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042694#comment-15042694
]
Phil Yang commented on HBASE-14004:
---
After discussion in HBASE-14790 , we can move forward now. Let me
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14999866#comment-14999866
]
Yu Li commented on HBASE-14004:
---
[~chenheng] [~Apache9]
Sorry for the late response, somehow didn't notice
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14977908#comment-14977908
]
Anoop Sam John commented on HBASE-14004:
In case when RS is killed and so later WAL replay will
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14977841#comment-14977841
]
Heng Chen commented on HBASE-14004:
---
{quote}
What if the regionserver crashed before flushing HFile? I
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14978108#comment-14978108
]
Duo Zhang commented on HBASE-14004:
---
[~carp84] This is an inherent problem of RPC based systems due to
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14978004#comment-14978004
]
Yu Li commented on HBASE-14004:
---
{quote}
It's possible that the HDFS sync RPC call fails, but the data is
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14978038#comment-14978038
]
Heng Chen commented on HBASE-14004:
---
On the assumption that HDFS sync successfully and do response,
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14977691#comment-14977691
]
Heng Chen commented on HBASE-14004:
---
I have a proposal.
When WAL sync failed, and master has to
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14977696#comment-14977696
]
Heng Chen commented on HBASE-14004:
---
{quote}
When WAL sync failed, and master has to rollback Memstore.
[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14977769#comment-14977769
]
Duo Zhang commented on HBASE-14004:
---
The problem here not only effects replication.
{quote}
As a
100 matches
Mail list logo