[
https://issues.apache.org/jira/browse/HBASE-3787?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13629812#comment-13629812
]
stack commented on HBASE-3787:
------------------------------
Patch looking good.
Add a comment that says what the result looks like when:
generateClientId
happens.
Why not just a UUID rather than all these gyrations? Or do you want to make it
so that looking at id, you can tell what client it came from? It looks like
you throw away all this info when you create the SecureRandom? Creating a
SecureRandom for this one time use is expensive.
Client id should be long since in proto is uint64 in proto?
Does ClientNonceManager have to be in top-level? Can it not be in client
package and be made package private?
Does it make sense putting clientid together w/ nonce making? Could you have a
class that does noncemaking and then another to hold the clientid? Is clientid
tied to Connection? Can you get connectionid? Or make a connectionid?
Connections are keyed by Configuration already? Would the Connection key do as
a clientid?
Would it be easier or make it so you could shut down access on
ClientNonceManager by passing in the id only rather than the whole nonce when
you do this:
MutateRequest request = RequestConverter.buildMutateRequest(
- location.getRegionInfo().getRegionName(), append);
+ location.getRegionInfo().getRegionName(), append, clientId,
nonce);
So, you decided to not pass nonce in here:
+ r = region.append(append, append.getWriteToWAL()/*, clientId2,
nonce*/);
I like the way this works over on the server side.
You dup code in append and increment.
Good stuff Sergey.
> Increment is non-idempotent but client retries RPC
> --------------------------------------------------
>
> Key: HBASE-3787
> URL: https://issues.apache.org/jira/browse/HBASE-3787
> Project: HBase
> Issue Type: Bug
> Components: Client
> Affects Versions: 0.94.4, 0.95.2
> Reporter: dhruba borthakur
> Assignee: Sergey Shelukhin
> Priority: Critical
> Fix For: 0.95.1
>
> Attachments: HBASE-3787-partial.patch
>
>
> The HTable.increment() operation is non-idempotent. The client retries the
> increment RPC a few times (as specified by configuration) before throwing an
> error to the application. This makes it possible that the same increment call
> be applied twice at the server.
> For increment operations, is it better to use
> HConnectionManager.getRegionServerWithoutRetries()? Another option would be
> to enhance the IPC module to make the RPC server correctly identify if the
> RPC is a retry attempt and handle accordingly.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
