[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13161486#comment-13161486
]
Rick Branson commented on CASSANDRA-3248:
-
#3411 should make this irrelevant
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13138101#comment-13138101
]
Peter Schuller commented on CASSANDRA-3248:
---
Empirically (but anecdotally),
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13136991#comment-13136991
]
Brandon Williams commented on CASSANDRA-3248:
-
ext3 w/fsync:
{noformat}
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137004#comment-13137004
]
Zhu Han commented on CASSANDRA-3248:
The log device is backed by no volatile RAM?
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137046#comment-13137046
]
Brandon Williams commented on CASSANDRA-3248:
-
No, plain SATA drives.
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137148#comment-13137148
]
Zhu Han commented on CASSANDRA-3248:
Is the write cache on the SATA drives turned
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137275#comment-13137275
]
Jason Rutherglen commented on CASSANDRA-3248:
-
Brandon, in the test, which
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137410#comment-13137410
]
Brandon Williams commented on CASSANDRA-3248:
-
bq. Is the write cache on
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137426#comment-13137426
]
Peter Schuller commented on CASSANDRA-3248:
---
XFS should detect write barrier
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137567#comment-13137567
]
Brandon Williams commented on CASSANDRA-3248:
-
I had a sw raid array
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137585#comment-13137585
]
Peter Schuller commented on CASSANDRA-3248:
---
Actually I think md does break
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137794#comment-13137794
]
Zhu Han commented on CASSANDRA-3248:
bq. XFS should detect write barrier support
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13137795#comment-13137795
]
Zhu Han commented on CASSANDRA-3248:
Depending on implementation, fsync might
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13126809#comment-13126809
]
Jonathan Ellis commented on CASSANDRA-3248:
---
Any news, Zhu?
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13127207#comment-13127207
]
Zhu Han commented on CASSANDRA-3248:
The developer has confirmed, on XFS, there is
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13114495#comment-13114495
]
Peter Schuller commented on CASSANDRA-3248:
---
@zhu Your initial benchmark in
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13114514#comment-13114514
]
Zhu Han commented on CASSANDRA-3248:
From xfs code, fsync() and fdatasync() should
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13113380#comment-13113380
]
Jonathan Ellis commented on CASSANDRA-3248:
---
File size is synced to disk by
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13113388#comment-13113388
]
Zhu Han commented on CASSANDRA-3248:
The manual said:
{quote}
$man fdatasync
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13113392#comment-13113392
]
Zhu Han commented on CASSANDRA-3248:
And postgresql uses fdatasync() to flush the
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13113397#comment-13113397
]
Jonathan Ellis commented on CASSANDRA-3248:
---
Postgresql can do that because
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13113437#comment-13113437
]
Zhu Han commented on CASSANDRA-3248:
I checked the source code of XFS which is
[
https://issues.apache.org/jira/browse/CASSANDRA-3248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13113466#comment-13113466
]
Zhu Han commented on CASSANDRA-3248:
It is very strange that there is no different
23 matches
Mail list logo
