[
https://issues.apache.org/jira/browse/CASSANDRA-13987?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16277811#comment-16277811
]
Jason Brown commented on CASSANDRA-13987:
-----------------------------------------
I've addressed the nits, and pushed to the usual spots:
||3.0||3.11||trunk||
|[branch|https://github.com/jasobrown/cassandra/tree/13987-3.0]|[branch|https://github.com/jasobrown/cassandra/tree/13987-3.11]|[branch|https://github.com/jasobrown/cassandra/tree/13987-trunk]|
|[utests &
dtests|https://circleci.com/gh/jasobrown/workflows/cassandra/tree/13987-3.0]|[utests
&
dtests|https://circleci.com/gh/jasobrown/workflows/cassandra/tree/13987-3.11]|[utests
&
dtests|https://circleci.com/gh/jasobrown/workflows/cassandra/tree/13987-trunk]|
||
bq. In trunk, the comment near the top of CLS::sync is missing a word.
I meant to remove this comment from trunk; already did it on 3.0/3.11
bq. Also in CLS::sync, why does the calculation of needToMarkData include
SYNC_MARKER_SIZE? I could be missing something, but it seems to me that we only
need check allocatePosition.get() > lastMarkerOffset
The [original
code|https://github.com/apache/cassandra/blob/trunk/src/java/org/apache/cassandra/db/commitlog/CommitLogSegment.java#L311]
did this (across all three branches). More properly, though, the call to
{{#allocate(int)}} returns the offset of where to begin writing into th\
e buffer, and that's what we store in {{lastMarkerOffset}}. So then on the next
time {{#sync()}} is invoked, we know another mutation has come in because
{{allocatePosition}} will be greater than {{lastMarkerOffset +
SYNC_MARKER_SIZE}}.
bq. dtests seem to be pretty broken on all 3 branches, is that teething
troubles with the custom circleci yaml
This is due to the circleci "teething troubles"
bq. On 3.0, in ACLS's runnable shouldn't this be checking shutdown not run
because we want to flush if shutdown has been requested?
D'oh! You are correct. fixed
> Multithreaded commitlog subtly changed durability
> -------------------------------------------------
>
> Key: CASSANDRA-13987
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13987
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Jason Brown
> Assignee: Jason Brown
> Fix For: 4.x
>
>
> When multithreaded commitlog was introduced in CASSANDRA-3578, we subtly
> changed the way that commitlog durability worked. Everything still gets
> written to an mmap file. However, not everything is replayable from the
> mmaped file after a process crash, in periodic mode.
> In brief, the reason this changesd is due to the chained markers that are
> required for the multithreaded commit log. At each msync, we wait for
> outstanding mutations to serialize into the commitlog, and update a marker
> before and after the commits that have accumluated since the last sync. With
> those markers, we can safely replay that section of the commitlog. Without
> the markers, we have no guarantee that the commits in that section were
> successfully written, thus we abandon those commits on replay.
> If you have correlated process failures of multiple nodes at "nearly" the
> same time (see ["There Is No
> Now"|http://queue.acm.org/detail.cfm?id=2745385]), it is possible to have
> data loss if none of the nodes msync the commitlog. For example, with RF=3,
> if quorum write succeeds on two nodes (and we acknowledge the write back to
> the client), and then the process on both nodes OOMs (say, due to reading the
> index for a 100GB partition), the write will be lost if neither process
> msync'ed the commitlog. More exactly, the commitlog cannot be fully replayed.
> The reason why this data is silently lost is due to the chained markers that
> were introduced with CASSANDRA-3578.
> The problem we are addressing with this ticket is incrementally improving
> 'durability' due to process crash, not host crash. (Note: operators should
> use batch mode to ensure greater durability, but batch mode in it's current
> implementation is a) borked, and b) will burn through, *very* rapidly, SSDs
> that don't have a non-volatile write cache sitting in front.)
> The current default for {{commitlog_sync_period_in_ms}} is 10 seconds, which
> means that a node could lose up to ten seconds of data due to process crash.
> The unfortunate thing is that the data is still avaialble, in the mmap file,
> but we can't replay it due to incomplete chained markers.
> ftr, I don't believe we've ever had a stated policy about commitlog
> durability wrt process crash. Pre-2.0 we naturally piggy-backed off the
> memory mapped file and the fact that every mutation was acquired a lock and
> wrote into the mmap buffer, and the ability to replay everything out of it
> came for free. With CASSANDRA-3578, that was subtly changed.
> Something [~jjirsa] pointed out to me is that [MySQL provides a way to adjust
> the durability
> guarantees|https://dev.mysql.com/doc/refman/5.6/en/innodb-parameters.html#sysvar_innodb_flush_log_at_trx_commit]
> of each commit in innodb via the {{innodb_flush_log_at_trx_commit}}. I'm
> using that idea as a loose springboard for what to do here.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
