ISPN-14207 Make MemcachedDecoder non blocking

[pruivo]

https://issues.redhat.com/browse/ISPN-14207

[pruivo]

looks like I have some failures in integration tests...

[wburns]

Looks good, just one real question regarding stats blocking.

[pruivo]

@diegolovison can we have a perf run with memcached client?

[wburns]

This makes me think we should add a test that does 2 operations in parallel with the non blocking mecached API. Or do we already have that?

[pruivo]

I can remove the get() from here: https://github.com/infinispan/infinispan/pull/10379/files#diff-49239ed285f72772402428f971aa797f4a8be001e4cf22c5a31765afe0e207ebR50

The test only needs the last get() to make sure the flush() is not executed concurrently with the puts.

[pruivo]

I did the change above ^

[wburns]

Only problem is that we aren't checking the response and they aren't interesting in the test. I would say we should do this to a test that does two gets in parallel that get different values in the response, just to make sure the responses aren't interleaved.

[pruivo]

added org.infinispan.server.functional.MemcachedOperations#testConcurrentGets

[wburns]

Great, thanks 👍

[wburns]

LGTM, waiting on CI/performance checks

[diegolovison]

http://perfrepo.mw.lab.eng.bos.redhat.com/reports/tableComparisonReport/19584

[pruivo]

@diegolovison are you sure the job did run the right thing?

• the snapshot commit is from 8 days ago.
• the logs show the test run for 240 sec but the report shows almost a ~400 sec duration.
• the radargun version used is different

[diegolovison]

http://perfrepo.mw.lab.eng.bos.redhat.com/reports/tableComparisonReport/19586

[tristantarrant]

Merged, thanks