Integration tests for the KryoNet communication

Description

Migrated from NovaTec Jira. Original ticket: INSPECTIT-1104

I would like to have an integration test which simulates on the side of the Agent the RMIConnection class and on the CMR the RegistrationStorage service. The test/simulation should validate that the serialization/sending/receiving/deserialization does not have any bottlenecks and bugs.

Environment

None

Activity

Show:
Ivan Senic
September 24, 2015, 12:46 PM

I give up for now.. Don't know what else to try out.. The main problem is that it's so hard to reproduce this.. It's not related to the data I am sending over, it's related to some specific kind of the situation that occurs.. And this happens not so often, I would say now it s ranging 1 out of 500 will fail..

It also seams that when I add debugging outputs, then it does not happen at all, as because of the additional code maybe we are not reaching the situation when it breaks..

I am making a break from this ticket. I invested more than 10 hours in total and had almost no progress..

Ivan Senic
October 8, 2015, 2:45 PM

I officially give up 100%.. I run app. 15,000 tests today and only two failed.. Without a way for reproducing, it's impossible to figure it our..

Btw. I created a checker that would confirm that sending/receiving was in order (same sequence of bytes written/read) and no test failed with the checker.. It can be due to the fact that I needed to change a bit the way of how I read the data.. Then I thought OK that's it, maybe reading is a problem, but then again seams like it's not, seams it's totally random.. I start to think that there is something random in this to make it fire up (like weather, position of the moon)..

Anyway totally disappointed.. I propose for sake of my health, lets not have this in next sprint..

Patrice Bouillet
October 9, 2015, 5:39 AM

Did you save the stream of bytes for the cases when the problem occurrs like we discussed? So that we can try to analyze it on that level and maybe even execute the same test with the same data again?

Ivan Senic
March 8, 2016, 3:17 PM

Finally

Ivan Senic
March 14, 2016, 2:48 PM
Edited

Seams like our friends from the CZB have hit the same bug. At least this is what I suspect. Thus I increase the priority:

Assignee

Ivan Senic

Reporter

Stefan Siegl

Labels

None

Integrator

Patrice Bouillet

Components

Sprint

None

Fix versions

Priority

Highest
Configure