Netvirt CSIT jobs are failing with Exception "Elasticsearch".
Srinivas <srinivas.rachakonda@...>
Hi All,
Test-netvirt-current CSIT jobs are failing due to issue/Exception with “Elasticsearch”.
Yesterday there was a commit in builder file: jjb/integration/common-functions.sh https://github.com/opendaylight/releng-builder/commit/a88663a066fff020a39b6b21ebdb5d520977e553
Could this be the cause of failure. Please kindly help.
Thanks, Srinivas +91-9243478719
|
|
Stanislav Chlebec <Stanislav.Chlebec@...>
Hi all I tried to analyze the problem. I think that it is due to the fact that patch https://git.opendaylight.org/gerrit/c/integration/test/+/85394 which removed the explicit mention of stuff related to JVM from scriptplan and testplan was merged on Nov 12 6:53 PM The patch https://git.opendaylight.org/gerrit/c/releng/builder/+/85398 Was merged on Nov 11 9:19 PM
And this build https://jenkins.opendaylight.org/releng/job/netvirt-csit-1node-0cmb-1ctl-2cmp-openstack-queens-gate-stateful-magnesium/118/ was executed on Nov 12, 2019 11:15:47 AM)
We well see if the next build is going to be successfull
Stan
From: srinivas.rachakonda@... <srinivas.rachakonda@...>
Hi All,
Test-netvirt-current CSIT jobs are failing due to issue/Exception with “Elasticsearch”.
Yesterday there was a commit in builder file: jjb/integration/common-functions.sh https://github.com/opendaylight/releng-builder/commit/a88663a066fff020a39b6b21ebdb5d520977e553
Could this be the cause of failure. Please kindly help.
Thanks, Srinivas +91-9243478719
|
|
Luis Gomez
Correct there is a time gap between 2 patches that can cause issues in some CSITs. Please revert if problem persists.
toggle quoted messageShow quoted text
|
|
Stanislav Chlebec <Stanislav.Chlebec@...>
Well I was wrong that was the reason. I tried to run the failing job without JVM support (patched with the new patch which I prepared: Setting the parameter ELASTICSEARCHATTRIBUTE=disabled Will disable all JVM related stuff in job) Here the job is without those failures So I conclude that there is really some connection between JVM support and those failures.
I also compared these two outputs for the same job: https://jenkins.opendaylight.org/sandbox/job/stan-fixed-netvirt-csit-1node-0cmb-1ctl-2cmp-openstack-queens-dcgw-sodium/1/ - with JVM support https://jenkins.opendaylight.org/sandbox/job/stanislav-netvirt-csit-1node-0cmb-1ctl-2cmp-openstack-queens-dcgw-sodium/2/ - without JVM support
Currently I do not know what is the reason for faulures or how to fix it... Stan
From: Luis Gomez <ecelgp@...>
Correct there is a time gap between 2 patches that can cause issues in some CSITs. Please revert if problem persists.
|
|
Srinivas <srinivas.rachakonda@...>
toggle quoted messageShow quoted text
From: Stanislav Chlebec <Stanislav.Chlebec@...> Sent: Thursday 14 November 2019, 12:37 PM To: Luis Gomez Cc: srinivas.rachakonda@... ; integration-dev@... Subject: RE: [integration-dev] Netvirt CSIT jobs are failing with Exception "Elasticsearch". Well I was wrong that was the reason. I tried to run the failing job without JVM support (patched with the new patch which I prepared: Setting the parameter ELASTICSEARCHATTRIBUTE=disabled Will disable all JVM related stuff in job) Here the job is without those failures So I conclude that there is really some connection between JVM support and those failures.
I also compared these two outputs for the same job: https://jenkins.opendaylight.org/sandbox/job/stan-fixed-netvirt-csit-1node-0cmb-1ctl-2cmp-openstack-queens-dcgw-sodium/1/ - with JVM support https://jenkins.opendaylight.org/sandbox/job/stanislav-netvirt-csit-1node-0cmb-1ctl-2cmp-openstack-queens-dcgw-sodium/2/ - without JVM support
Currently I do not know what is the reason for faulures or how to fix it... Stan
From: Luis Gomez <ecelgp@...>
Correct there is a time gap between 2 patches that can cause issues in some CSITs. Please revert if problem persists.
|
|
Stanislav Chlebec <Stanislav.Chlebec@...>
This is one possibility.
Can someone more experienced than me to see this recomendation about elasticsearch: Elasticsearch settings are in $ES_HOME/config/elasticsearch.yml. There, if the cluster.name setting is commented out, it means ES would take just about any cluster name. So, in your code, the cluster.name as "elastictest" might be the problem. Try this: Client client = new TransportClient() .addTransportAddress(new InetSocketTransportAddress( "143.79.236.xxx", 9300));
Maybe it is also our problem.... Stan
From: Srinivas Rachakonda <srinivas.rachakonda@...>
Hi,
Can the changes be reverted back as many CSIT jobs are failing.
From: Stanislav Chlebec <Stanislav.Chlebec@...>
Well I was wrong that was the reason. I tried to run the failing job without JVM support (patched with the new patch which I prepared: Setting the parameter ELASTICSEARCHATTRIBUTE=disabled Will disable all JVM related stuff in job) Here the job is without those failures So I conclude that there is really some connection between JVM support and those failures.
I also compared these two outputs for the same job: https://jenkins.opendaylight.org/sandbox/job/stan-fixed-netvirt-csit-1node-0cmb-1ctl-2cmp-openstack-queens-dcgw-sodium/1/ - with JVM support https://jenkins.opendaylight.org/sandbox/job/stanislav-netvirt-csit-1node-0cmb-1ctl-2cmp-openstack-queens-dcgw-sodium/2/ - without JVM support
Currently I do not know what is the reason for faulures or how to fix it... Stan
From: Luis Gomez <ecelgp@...>
Correct there is a time gap between 2 patches that can cause issues in some CSITs. Please revert if problem persists.
|
|
Luis Gomez
The real issue is apache decanter does not work in the latest Karaf version pulled in master. This generates an exception in the Karaf log and it seems netvirt tests are tracking and failing any test with karaf exceptions, so the solution is to merge this patch [1] and disable the framework in all netvirt jobs in master or in all netvirt jobs if netvirt project is not interested in JVM monitoring.
toggle quoted messageShow quoted text
|
|
JamO Luhrsen
or you can add the exception to this list:
toggle quoted messageShow quoted text
https://github.com/opendaylight/integration-test/blob/master/csit/libraries/netvirt/excepts.py eventually, assuming magnesium starts to work with decanter, you will hopefully get the jvm monitoring. JamO On 11/14/19 9:44 AM, Luis Gomez wrote:
The real issue is apache decanter does not work in the latest Karaf version pulled in master. This generates an exception in the Karaf log and it seems netvirt tests are tracking and failing any test with karaf exceptions, so the solution is to merge this patch [1] and disable the framework in all netvirt jobs in master or in all netvirt jobs if netvirt project is not interested in JVM monitoring.
|
|
Luis Gomez
Thanks Jamo, I guess if netvirt project is interested in the JVM framework they can patch as you suggest, otherwise we can just disable it for netvirt so we have less things to maintain in integration.
toggle quoted messageShow quoted text
BR/Luis
|
|
Luis Gomez
Also looping netvirt list, to see how they want to proceed here.
toggle quoted messageShow quoted text
|
|