Affects Version/s: 7.0.0
Fix Version/s: 7.0.0
Environment:IDM 7.0.0 master build runs on a CentOS 7 machine. Java 11.
We see significant recon throughput degradation in some of our MySQL based recon tests, one test is recon from Managed to DJ:
Before the changes from
OPENIDM-13101, in commit 69fdc1d1a, we have: (Duration is in second and tps is throughput)
OPENIDM-13101(commit 5d0e3fe), we have:
It's a significant drop.
The symptom is seen in recon from IDM to DJ tests, and from DJ to IDM recon update tests(didn't see it in recon from DJ to IDM create tests though) and didn't see it in DJ based repo tests.
The symptom can be seen from the very beginning of the test, for a long time after recon starts, little/no progress were made:
To reproduce using PyForge:
Change stress parameters as follow:
and OpenIDM parameters:
and run the test as: