[Hawkular-dev] Metrics: Cassandra Driver timeouts during the build

[Peter Palaga]

Ahoj Filip, inline...

On 2016-05-31 14:18, Filip Brychta wrote:
> I've seen this exception several times but only when hitting h-metrics with heavy load -> if there is no load it will be probably env issue.

Any ideas which part of the env should I check?

> I also noticed that it takes longer to start up cassandra version 3.x then 2.x. Is the cassandra node fully up before starting h-metrics?

Yes, I have run ccm start in the morning and the builds are failing 
constantly during the whole day.

-- P

> Filip
>
>
> ----- Original Message -----
>> Hi *,
>>
>> Short version: doubling the read timeouts [1] for all three uses of
>> Cassandra Driver solved the problem for me. Is this a good idea? Can
>> anybody see a better solution to my problem?
>>
>>
>> Long version:
>>
>> https://issues.jboss.org/browse/HWKMETRICS-404
>>
>> As some of you already know from my complaints on IRC, I have not been
>> able to build Metrics since a couple of weeks.
>>
>> My builds failed mostly on core-services but sometimes also in
>> configuration-service and resr-tests-jaxrs, the error always being the same:
>>
>> com.datastax.driver.core.exceptions.NoHostAvailableException: All
>> host(s) tried for query failed (tried: /127.0.0.1:9042
>> (com.datastax.driver.core.exceptions.OperationTimedOutException:
>> [/127.0.0.1] Timed out waiting for server response))
>>
>> The whole build log: http://paste.fedoraproject.org/372623/46116551/
>>
>> This started happening at some point after my PR
>> https://github.com/hawkular/hawkular-metrics/pull/478 was merged. I do
>> not say this PR is the cause, this is just an attempt to narrow down the
>> set of commits that might have introduced the problem.
>>
>> As nobody else (incl. Travis) complained about anything similar, I was
>> thinking the issue must be somewhere in my environment. I double checked
>> I use the right Cassandra version, ulimits, Java version, Maven version,
>> system updates, but nothing helped.
>>
>> Then I tried doubling the read timeouts [1] for all three uses of
>> Cassandra Driver and the build started passing again. Is this a good
>> idea? Can anybody see a better solution to my problem?
>>
>> [1] https://github.com/hawkular/hawkular-metrics/pull/510
>> _______________________________________________
>> hawkular-dev mailing list
>> hawkular-dev at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/hawkular-dev
>>
> _______________________________________________
> hawkular-dev mailing list
> hawkular-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hawkular-dev
>