[
https://issues.apache.org/jira/browse/TEPHRA-194?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15593282#comment-15593282
]
ASF GitHub Bot commented on TEPHRA-194:
---------------------------------------
Github user anew commented on a diff in the pull request:
https://github.com/apache/incubator-tephra/pull/18#discussion_r84392263
--- Diff:
tephra-core/src/main/java/org/apache/tephra/distributed/TransactionServiceThriftClient.java
---
@@ -112,7 +113,20 @@ public Transaction startShort() throws TException {
public Transaction startShort(int timeout) throws TException {
try {
- return
TransactionConverterUtils.unwrap(client.startShortTimeout(timeout));
+ return
TransactionConverterUtils.unwrap(client.startShortWithTimeout(timeout));
+ } catch (TGenericException e) {
+ isValid.set(false);
+ // currently, we only expect IllegalArgumentException here, if the
timeout is invalid
+ IllegalArgumentException actual;
+ try {
+ actual = (IllegalArgumentException)
+ Class.forName(e.getOriginalExceptionClass())
+ .getConstructor(String.class)
+ .newInstance(e.getMessage());
+ } catch (Throwable t) {
--- End diff --
Because we do not expect anything else than IllegalArgumentException? For
now, I will change the code to be more readable and expect exactly that.
> Transaction client should not retry startShort() if an invalid timeout is
> given
> --------------------------------------------------------------------------------
>
> Key: TEPHRA-194
> URL: https://issues.apache.org/jira/browse/TEPHRA-194
> Project: Tephra
> Issue Type: Bug
> Components: client
> Affects Versions: 0.9.0-incubating, 0.10.0-incubating
> Reporter: Andreas Neumann
> Assignee: Andreas Neumann
> Fix For: 0.10.0-incubating
>
>
> Currently, if an invalid timeout (negative, or too long) is given, the Tx
> manager throws an IllegalArgumentException. The thrift client will catch that
> and apply the retry strategy. However, in this case, retry is pointless, and
> if the strategy is, for example, exponential backoff, if introduces
> unneccessary load and latency.
> The service should instead throw a meaningful exception for that, so that the
> client knows not to retry.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
