[
https://issues.apache.org/jira/browse/QPID-7427?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alex Rudyy updated QPID-7427:
-----------------------------
Description:
Test MultiNodeTest.testInFlightTransactionsWhilstMajorityIsLost can fail
sporadically on sending REST request to activate the stopped node which has
transactions ahead of master causing the Environment to restart.
On activation the JE environment is created. If it detects ahead of master
transactions it restarts itself. The activation logic within BDB VHN accesses
the Environment via EnvironmentFacade but "ConnectionScopedRuntimeException:
Environment is restarting" is thrown causing VHN transition into ERRORED state.
REST request ends up in 500 http status code.
The operator would be required to repeat the REST request in order to finish
activation. However, 500 status code can potentially confuse the end user.
Additionally activation logic could be retried inside of activate method on
running into "Environment restarting" exception.
The failed test (MultiNodeTest.testInFlightTransactionsWhilstMajorityIsLos) on
detection of 500 status code just logs it and ignore. It seems that test
optimistically assumes that activation will be always successful.We need to fix
a test to repeat the activation if it fails.
was:
Test MultiNodeTest.testInFlightTransactionsWhilstMajorityIsLost can fail
sporadically on sending REST request to activate the stopped node which has
transactions ahead of master causing the Environment to restart.
On activation the JE environment is created. If it detects ahead of master
transactions it restarts itself. The activation logic within BDB VHN accesses
the Environment via EnvironmentFacade but "ConnectionScopedRuntimeException:
Environment is restarting" is thrown causing VHN transition into ERRORED state.
REST request ends up in 500 http status code.
The operator would be required to repeat the REST request in order to finish
activation. However, 500 status code can potentially confuse the end user.
Additionally activation logic could be retried inside of activate mode on
running into "Environment restarting" exception.
The failed test (MultiNodeTest.testInFlightTransactionsWhilstMajorityIsLos) on
detection of 500 status code just logs it and ignore. It seems that test
optimistically assumes that activation will be always successful.We need to fix
a test to repeat the activation if it fails.
> [Java Broker, HA] An activation of stopped node might fail when node has
> ahead of master transactions causing Environment to restart
> ------------------------------------------------------------------------------------------------------------------------------------
>
> Key: QPID-7427
> URL: https://issues.apache.org/jira/browse/QPID-7427
> Project: Qpid
> Issue Type: Bug
> Components: Java Broker
> Affects Versions: 0.32, qpid-java-6.0, qpid-java-6.0.1, qpid-java-6.0.2,
> qpid-java-6.0.3, qpid-java-6.0.4, qpid-java-6.1
> Reporter: Alex Rudyy
>
> Test MultiNodeTest.testInFlightTransactionsWhilstMajorityIsLost can fail
> sporadically on sending REST request to activate the stopped node which has
> transactions ahead of master causing the Environment to restart.
> On activation the JE environment is created. If it detects ahead of master
> transactions it restarts itself. The activation logic within BDB VHN accesses
> the Environment via EnvironmentFacade but "ConnectionScopedRuntimeException:
> Environment is restarting" is thrown causing VHN transition into ERRORED
> state. REST request ends up in 500 http status code.
> The operator would be required to repeat the REST request in order to finish
> activation. However, 500 status code can potentially confuse the end user.
> Additionally activation logic could be retried inside of activate method on
> running into "Environment restarting" exception.
> The failed test (MultiNodeTest.testInFlightTransactionsWhilstMajorityIsLos)
> on detection of 500 status code just logs it and ignore. It seems that test
> optimistically assumes that activation will be always successful.We need to
> fix a test to repeat the activation if it fails.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
