[
https://issues.apache.org/jira/browse/CASSANDRA-9328?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14980613#comment-14980613
]
Jim Meyer commented on CASSANDRA-9328:
--------------------------------------
I'm not sure I understand the proposed workaround. For example, suppose two
clients are trying to create the same username in a table. If one of them gets
a WTE, then they can't do a simple read to see if their insert was successful
since the other client may have been the one that created it.
So it seems that each client would need to set the data in a unique way, such
that it can do a simple read and determine that its specific transaction was
applied. For example, the client could set a UUID field as a transaction id,
and then on the extra read, check if the UUID matched what they wrote to
differentiate their LWT from those of other clients. I guess that would work,
but besides being slow, it would waste quite a bit of space to store
transaction ID's. Is there a better way?
> WriteTimeoutException thrown when LWT concurrency > 1, despite the query
> duration taking MUCH less than cas_contention_timeout_in_ms
> ------------------------------------------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-9328
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9328
> Project: Cassandra
> Issue Type: Bug
> Reporter: Aaron Whiteside
> Priority: Critical
> Fix For: 2.1.x
>
> Attachments: CassandraLWTTest.java, CassandraLWTTest2.java
>
>
> WriteTimeoutException thrown when LWT concurrency > 1, despite the query
> duration taking MUCH less than cas_contention_timeout_in_ms.
> Unit test attached, run against a 3 node cluster running 2.1.5.
> If you reduce the threadCount to 1, you never see a WriteTimeoutException. If
> the WTE is due to not being able to communicate with other nodes, why does
> the concurrency >1 cause inter-node communication to fail?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
