[
https://issues.apache.org/jira/browse/JCR-3979?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15291164#comment-15291164
]
Vikas Saurabh commented on JCR-3979:
------------------------------------
There can a few options to do a better job at handling such cases:
# We can have retry/upload and maxNumOfFailures (both configurable) to reduce
the probability of initial sync failue.
## If maxNumOfFailures aren't hit then we can retry those uploads which failed
at the end of initial sync to have a go at them again
# Don't fail sync on upload failure, but keep track of uploads that failed.
Then, as phase 2 of initial sync, retry those failures
# Identify the cause of upload failure and do some back-off in case the error
is more to do with choking the capabilities (over zealous configurations)
(we can, of course, do a mix of any of those)
(cc [~shgupta], [~amjain])
> Improve resilience when s3 is being initialized by local datastore
> ------------------------------------------------------------------
>
> Key: JCR-3979
> URL: https://issues.apache.org/jira/browse/JCR-3979
> Project: Jackrabbit Content Repository
> Issue Type: Bug
> Components: jackrabbit-data
> Reporter: Vikas Saurabh
> Assignee: Vikas Saurabh
> Priority: Minor
>
> Currently, during initialization, any upload error on any thread causes
> initialization to fail. This makes initialization of big repositories quite
> painful.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
