[Hawkular-dev] Could not connect to Cassandra ... - does it have to be a WARN with a stack trace?

[Matt Wringe]

There are a couple of things here.

One is the embedded Cassandra. I don't know if its a dependency issue in 
that we are not defining the embedded Cassandra as a dependency properly 
or if its something along the lines of that the dependency is started, 
but not yet able to accept connections yet. Either way, it looks like we 
need to figure that out and handle the warning in the logs better (which 
could be fixing the dependency management, or making it info level for 
the first couple of failures).

The other issue is when we are using an external Cassandra and its not 
yet available to take connections when Hawkular is started. In Hawkular 
Metrics anyways, it will start and throw the warning messages in the 
logs until Cassandra is fully running. This is done so that we can start 
Hawkular Metrics and Cassandra at the same time and not have to worry 
about the order of things. This is really handy if Cassandra and 
Hawkular Metrics are configured as services to start when the machine is 
rebooted, and also when dealing with containers (especially since 
containers don't have the same concept of dependencies).

But maybe these should be changed to 'info' level for the first while 
and only then transformed into 'warnings'. Or handled in some other 
manner. Its a bit weird since when starting containers at the same time, 
the initial failures are expected, but if they persist, it usually means 
the Cassandra container was not started or failed to startup properly.

On 27/07/15 05:55 AM, Lucas Ponce wrote:
> +1
>
> The WARNING message is correct as the component cannot connect to Cassandra, so, I don't remove those messages.
>
> But as TH says, it's a matter of how to define the embedded cassandra in this scenario to be up and running before that the components.
>
> Lucas
>
> ----- Original Message -----
>> From: "Thomas Heute" <theute at redhat.com>
>> To: "Discussions around Hawkular development" <hawkular-dev at lists.jboss.org>
>> Sent: Monday, July 27, 2015 11:42:04 AM
>> Subject: Re: [Hawkular-dev] Could not connect to Cassandra ... - does it have to be a WARN with a stack trace?
>>
>> Glad I am not the only one that finds those messages annoying :)
>>
>> Indeed, this is an Hawkular issue with embedded Cassandra.
>> We should fix it by properly defining the startup dependencies, we may also
>> think of keeping "embedded Cassandra" as a separate process.
>>
>> The idea of "embedding Cassandra" is that we don't need to ask users to go
>> download it, "install it" and run it.
>> But we could consider having a Cassandra in a different process (configured
>> for our needs such as the data location, port and other settings) that is
>> required to be started prior to run Hawkular (Hawkular would need to fail
>> quickly and clearly if the user "forgot" to start Cassandra).
>>
>> PS: I am not advocating for one or the other solution as I don't know the
>> advantages/drawbacks of both solutions
>>
>> Thomas
>>
>>
>>
>>
>>
>> On 07/27/2015 11:11 AM, Michael Burman wrote:
>>> Hi,
>>>
>>> And how would you distinguish between real warning and incorrect one? This
>>> message is not printed when Hawkular-Metrics starts standalone andshould
>>> not be removed. This is a container issue in Hawkular and should be fixed
>>> there (so that no such error happens) instead of breaking the logging in
>>> metrics. Most likely this happens with embedded-cassandra I guess?
>>>
>>>     - Micke
>>>
>>> ----- Original Message -----
>>> From: "Peter Palaga" <ppalaga at redhat.com>
>>> To: "Discussions around Hawkular development"
>>> <hawkular-dev at lists.jboss.org>
>>> Sent: Monday, July 27, 2015 11:59:27 AM
>>> Subject: [Hawkular-dev] Could not connect to Cassandra ... - does it have
>>> to be a WARN with a stack trace?
>>>
>>> Hi *,
>>>
>>> There are several occurrences of this in every Hawkular start log:
>>>
>>> WARN  [org.hawkular.metrics.api.jaxrs.MetricsServiceLifecycle]
>>> (metricsservice-lifecycle-thread) Could not connect to Cassandra cluster
>>> - assuming its not up yet:
>>> com.datastax.driver.core.exceptions.NoHostAvailableException: All
>>> host(s) tried for query failed (tried: /127.0.0.1:9042
>>> (com.datastax.driver.core.TransportException: [/127.0.0.1:9042] Cannot
>>> connect))
>>>
>>> plus the stack trace.
>>>
>>> So given that this happens during every HK startup, could we not
>>> classify it as normal and change it to INFO without the stack trace?
>>>
>>> I am ready to prepare a PR unless somebody raises a hand against that.
>>>
>>> Thanks,
>>>
>>> Peter
>>> _______________________________________________
>>> 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
>>>
>> _______________________________________________
>> 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