[
https://issues.apache.org/jira/browse/HBASE-3293?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12966233#action_12966233
]
stack commented on HBASE-3293:
------------------------------
My guess would be that allowing negative timestamps would be more work than
rejecting them.
Regards times before 1970, that'd be a strange notion I'd say. Did anything of
any significance happen before 1970? (If I had my way, I'd have started the
clock after the 80s!).
> Negative timestamps can be inserted without fail but not read again
> -------------------------------------------------------------------
>
> Key: HBASE-3293
> URL: https://issues.apache.org/jira/browse/HBASE-3293
> Project: HBase
> Issue Type: Bug
> Environment: CentOS 5.5 with Cloudera's CDH3b3
> Reporter: fnord 999
>
> When inserting a value into HBase that has a negative timestamp, such as -1,
> HBase inserts that value without an exception or any kind of other warning if
> we use the Java API. The value is written to a region and the webinterface of
> the regionservers shows that there is data in them.
> However: neither a Scan nor a Get nor any other operation via MapReduce in
> the Shell or via Java can read that specific value out.
> There would be 2 different fixes in my opinion:
> 1. simply accept the negative timestamps consequently in read and write
> operations
> 2. if negative timestamps cannot be used, throw an exception when someone
> tries to insert an invalid timestamp
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
