[
https://issues.apache.org/jira/browse/PHOENIX-5718?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17031828#comment-17031828
]
Chinmay Kulkarni commented on PHOENIX-5718:
-------------------------------------------
This does not look like expected behavior. I'm surprised this never showed up
during any previous upgrade. Is this specific to 4.16 since the extra columns
for view TTL are not part of the upgrade path and thus this gets triggered from
the failure retry condition [~sandeep.guggilam]?
> GetTable builds a table excluding the given clientTimeStamp
> -----------------------------------------------------------
>
> Key: PHOENIX-5718
> URL: https://issues.apache.org/jira/browse/PHOENIX-5718
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.16.0
> Reporter: Sandeep Guggilam
> Assignee: Sandeep Guggilam
> Priority: Major
> Fix For: 4.16.0
>
>
> Here is the scenario tested:
> # Brought up a server with 4.16 where new columns are added but not added as
> part of upgrade path
> # Connect with 4.14 client
> # Connect with a 4.16 client - this will throw an exception as the new
> columns added as part of 4.16 were not added as part of upgrade path
> # Now the code will force update the cache in
> PhoenixStatement#executeQuery() method
> # Now the buildTable is removing even the columns added as part of 4.15 ,
> the reason being we are passing the clientTimeStamp to build table ( say 29
> is the timestamp for column added for 4.15) but the table is scanning rows
> EXCLUDING the passed clientTimeSTamp as the Scan#setTimeRange method excludes
> the end time stamp
> The passing of clientTimeStamp to build table is in
> MetaDataEndPointImpl#doGetTable method
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
