| Re: [kapua-dev] Build failures |
Sound good, so we will add new tests to check the subset filtering and especially the alignment between metrics mappings (in the metric registry) and the metrics contained in the messages
Il giorno 26 apr 2017, alle ore 17:51, Jens Reimann <jreimann@xxxxxxxxxx> ha scritto:
So if you want to actually test subsets of data, then I agree, we should add specific tests for that, testing all different ways to filter out data.
______________________________
On Wed, Apr 26, 2017 at 5:44 PM, Modanese, Riccardo <Riccardo.Modanese@eurotech.com > wrote:
I agree about the synchronization between insert and next querying but regarding the indices clearing I’m worried about make tests too simple, I mean, by clearing the indices every time may we can not find issues related to wrong registry synchronization.In any case we can proceed in that way and add more tests to cover this cases in the a next step, if you agree...
Il giorno 26 apr 2017, alle ore 17:29, Jens Reimann <jreimann@xxxxxxxxxx> ha scritto:
I do think that by clearing out the indices and the using the refresh method we now have clean state and the necessary synchronization. Which would make this independent of any machine specific timeout.Well the tests are actually working fine now. But I could be that non waiting for those 10 seconds will of course no introduce this implicit 2 second gap, which the date range was being used for. However this is not a problem between insert and wait, as this is taken care of by the refresh call. This is more a problem between unit test "a" and unit test "b".As now there the indices are cleared and the refresh is used, I see no reason to do any more waiting or date range querying, as this seemed to be problematic on different kinds of machine.
______________________________
On Wed, Apr 26, 2017 at 4:54 PM, Modanese, Riccardo <Riccardo.Modanese@xxxxxxxxxxxm > wrote:
I cannot figure out that since I’m working on es test since few weeks and usually I run several times the whole test suite without any failure.My current branch is quite different from develop but, if you take a look to the code, you can see that every test set up a different date for the message and also the channels are ofter differents.May the issue you are experienced was introduced by your fix (https://github.com/eclipse/kapua/commit/b926e2b97d7cda71bec )?3ed182cb511af5e51ad95 Since you removed the wait time between insert and query steps, now the tests are messed up! so to fix them the correct way is to set a different prefixed date for every test.
Il giorno 26 apr 2017, alle ore 16:44, Jens Reimann <jreimann@xxxxxxxxxx> ha scritto:
Let's keep this running for a while and see what happens.Even now on hudson it seems to work fine after removing all timeouts, delays and stuff.You can try this out locally. If you remove the "clear indices" method, the methods will find messages from other tests. Running only a single test works repeatedly, while running all test fails repeatedly with the same error message.This popped up after refreshing the indices, as then every query would have the actual set of data which got inserted. I guess before that change it was more like randomness which messages actually got returned in time.
______________________________
On Wed, Apr 26, 2017 at 4:23 PM, Modanese, Riccardo <Riccardo.Modanese@xxxxxxxxxxxm > wrote:
Hi,
it’s NOT true since the datastore unit tests are searching for messages in the specific channel and for a specific date range. So, even if all the test are running concurrently, every test should find only the right messages.Moreover some failures are due to a timeout occurring in the store phase. So it means that es instance it’s not able to store a message after 60 seconds!
Il giorno 26 apr 2017, alle ore 16:18, Morson, Stefano <stefano.morson@xxxxxxxxxxxx> ha scritto:
Yeah, cleaning the indexes before the tests is the right thing to do.
Thanks,Stefano
From: kapua-dev-bounces@eclipse.org <kapua-dev-bounces@eclipse.org > on behalf of Jens Reimann <jreimann@xxxxxxxxxx>
Sent: Wednesday, April 26, 2017 3:27:27 PM
To: kapua developer discussions
Subject: Re: [kapua-dev] Build failures______________________________JensCheersSo I would suggest to re-enable this flag in the near future.I fixed this by clearing out all indices before each unit test and now it works. This fixed it locally, on travis and should also work on hudson.I think I fixed most of those issues with the datastore tests.Beside the index "refresh" issue, which I mentioned and added there was some assumptions that the data found in those indices actually is only coming from the unit test in question. However running one unit test after each other filled up the storage an messed up the other tests. So running a single unit test works, but running all of them doesn't. Adding the refresh call actually broke it locally and on hudson, as it seems to me those tests never really worked when running batched.
On Wed, Apr 26, 2017 at 3:11 PM, Morson, Stefano <stefano.morson@eurotech.com > wrote:
Hi all,
I temporarily added the property
-Dmaven.test.failure.ignore=true
to the Hudson build for Kapua. This is because there are tests (especially the datastore tests) that are recurrently failing in Hudson but are perfectly working on more than one the local machine. This could be due to the amount of available resources in Hudson. Uros will investigate more the resource usage.
Using the flag above we still have the evidence that something went wrong without invalidating the whole build. As soon as we'll have a remedy the flag will be removed.
Stefano
From: kapua-dev-bounces@eclipse.org <kapua-dev-bounces@eclipse.org > on behalf of Morson, Stefano <stefano.morson@xxxxxxxxxxxx>
Sent: Tuesday, April 25, 2017 5:04:06 PM
To: kapua developer discussions
Subject: [kapua-dev] Build failuresHi all,
One of the issues we are having for the builds in these days was due to Sonar. I dropped an email to webmaster and he explained me this was due to a configuration change in the backends. He also fixed the issue in Kapua Hudson account so now we should be ok with this.
Stefano
Stefano Morson | EUROTECH | tel.: +39 0433 485411
_______________________________________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
--
Jens Reimann
Senior Software Engineer / EMEA ENG Middleware
Werner-von-Siemens-Ring 14
85630 Grasbrunn
Germany
phone: +49 89 2050 71286
____________________________________________________________ _________________
Red Hat GmbH, www.de.redhat.com,
Registered seat: Grasbrunn, Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Paul Argiry, Charles Cachera, Michael Cunningham, Michael O'Neill
_________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
_______________________________________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
--
Jens Reimann
Senior Software Engineer / EMEA ENG Middleware
Werner-von-Siemens-Ring 14
85630 Grasbrunn
Germany
phone: +49 89 2050 71286
____________________________________________________________ _________________
Red Hat GmbH, www.de.redhat.com,
Registered seat: Grasbrunn, Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Paul Argiry, Charles Cachera, Michael Cunningham, Michael O'Neill
_________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
_______________________________________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
--
Jens Reimann
Senior Software Engineer / EMEA ENG Middleware
Werner-von-Siemens-Ring 14
85630 Grasbrunn
Germany
phone: +49 89 2050 71286
____________________________________________________________ _________________
Red Hat GmbH, www.de.redhat.com,
Registered seat: Grasbrunn, Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Paul Argiry, Charles Cachera, Michael Cunningham, Michael O'Neill
_________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
_______________________________________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
--
Jens Reimann
Senior Software Engineer / EMEA ENG Middleware
Werner-von-Siemens-Ring 14
85630 Grasbrunn
Germany
phone: +49 89 2050 71286
____________________________________________________________ _________________
Red Hat GmbH, www.de.redhat.com,
Registered seat: Grasbrunn, Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Paul Argiry, Charles Cachera, Michael Cunningham, Michael O'Neill
_________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev
_______________________________________________
kapua-dev mailing list
kapua-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/kapua-dev