[
https://issues.apache.org/jira/browse/HADOOP-13761?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16356378#comment-16356378
]
Aaron Fabbri commented on HADOOP-13761:
---------------------------------------
Thanks for the feedback on the scale test patch [~ste...@apache.org]. On the
retries, I've been looking at open(), delete(), and rename() in terms of what
happens when MetadataStore has metadata but we hit eventual consistency when we
start operating on S3.
* open(): This is the only code that creates a S3AInputStream constructor, and
it does a getFileStatus() first, so failures should be happening later in
read() paths. There is already a single handler for read failures:
{{onReadFailure()}}. I suggest we add invoker/retry policy to this function.
Thoughts? As is, it appears to only retry once without any backoff timer.
* delete(): this uses deleteObject(), which already has retries. I think we
are good here?
* rename(): this is more complicated, and we probably need to do more in-depth
rename() rewrite to include HADOOP-15183. One thing I noticed is,
{{copyFile()}} is annotated {{RetryMixed}} but only uses {{once()}} internally,
so I might fix that.
I'd like to address Steve's feedback on the scale test and make the above
changes if it sounds good. I think rename() would benefit from its own
patch/JIRA, (possibly even add a new experimental "v2" rename codepath that can
be switched in via config while we improve and test with it, dunno).
> S3Guard: implement retries for DDB failures and throttling; translate
> exceptions
> --------------------------------------------------------------------------------
>
> Key: HADOOP-13761
> URL: https://issues.apache.org/jira/browse/HADOOP-13761
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 3.0.0-beta1
> Reporter: Aaron Fabbri
> Assignee: Aaron Fabbri
> Priority: Blocker
> Attachments: HADOOP-13761.001.patch, HADOOP-13761.002.patch
>
>
> Following the S3AFileSystem integration patch in HADOOP-13651, we need to add
> retry logic.
> In HADOOP-13651, I added TODO comments in most of the places retry loops are
> needed, including:
> - open(path). If MetadataStore reflects recent create/move of file path, but
> we fail to read it from S3, retry.
> - delete(path). If deleteObject() on S3 fails, but MetadataStore shows the
> file exists, retry.
> - rename(src,dest). If source path is not visible in S3 yet, retry.
> - listFiles(). Skip for now. Not currently implemented in S3Guard. I will
> create a separate JIRA for this as it will likely require interface changes
> (i.e. prefix or subtree scan).
> We may miss some cases initially and we should do failure injection testing
> to make sure we're covered. Failure injection tests can be a separate JIRA
> to make this easier to review.
> We also need basic configuration parameters around retry policy. There
> should be a way to specify maximum retry duration, as some applications would
> prefer to receive an error eventually, than waiting indefinitely. We should
> also be keeping statistics when inconsistency is detected and we enter a
> retry loop.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org
