[
https://issues.apache.org/jira/browse/FALCON-238?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13883941#comment-13883941
]
Samarth Gupta commented on FALCON-238:
--------------------------------------
Following test will be added to regression to test the above :
below values will be passed as "end" parameter to api
1. empty
2. random
3. ill formatted time stamp
4. update time in past
5. update time in next new mins
6. update time clash with new instance materialization
7. update time after process end time
8. update time before process start time when process start is in future
9. update when diff cluster have different validity , valid for some , invalid
others
10. above for both process and feed
11. update by adding new cluster where update is valid / invalid
12. update at the same time as process / feed start / end time when they are in
future
13. update when process / feed end was also in past.
will add more ....
> Support updates at specific time
> --------------------------------
>
> Key: FALCON-238
> URL: https://issues.apache.org/jira/browse/FALCON-238
> Project: Falcon
> Issue Type: New Feature
> Reporter: Shwetha G S
> Assignee: Shwetha G S
> Fix For: 0.5
>
> Attachments: FALCON-238-v2.patch, FALCON-238.patch
>
>
> Currently, feed/process update sets end time of coordinator to now and then
> creates new coord with start time as now. When updating a process's
> workflow/lib, its useful to stage an update with specific timestamp so that
> all instances before the timestamp use the old lib and instances after the
> timestamp use new lib. This is useful to roll-out any application changes
> This can be implemented by taking additional timestamp argument in update and
> honour that whenever possible. For cases where this can't be honoured, return
> the update timestamp always so that user is aware of when the new lib is used
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
