Details
-
Task
-
Resolution: Fixed
-
Major
-
None
Description
The major difference in between test runs on my little performance cluster and the large iris cluster is the impact that garbage collector load has.
It seems that on larger machines garbage collecting is more expensive by about 50%.
While request allocation is not the worse offender, it none the less shows up, and this is a simple fix to apply.
Attachments
For Gerrit Dashboard: MB-31620 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
100513,2 | MB-31620 avoid GC load on http request | master | query | Status: MERGED | +2 | +1 |
101745,3 | MB-31620 improve GC load on http request creation | master | query | Status: MERGED | +2 | +1 |