Antti Kervinen
Antti Kervinen
@marcemq, @grahamwhaley , do you think we could use `TEST_NAME` so that test scripts would save their output files directly under `results/TEST_NAME` subdirectory? That is, every test script would create...
Good points, thanks @grahamwhaley. I added the memory used plot. Now the page looks more complete, and the title offered a nice place to explain what "used" actually means. (That...
rebased, commits squashed, tested, ready for merge.
Fixed in PR #212
Brainstorming: as we have only few nodes in scaling clusters, do you think we'd use the role of the node instead of hostname of the node in the report? For...
Request for comments: - The default latency percentiles are now 5-25-50-75-95. - This means we visualize - median (50) - the range including 50 % of the measurements (25, 75)...
@grahamwhaley , I agree that millisecond granularity is rather coarse for reporting network latency. I get 12 ms on my VM cluster, but 3-4 ms on real hardware. However, I...
@grahamwhaley , good point. I changed the time in the result json to microsecond accuracy. We can handle the measurement method as another topic. In any case, current round trip...
Rebased the whole series. Now this is based on the rapid test. Quite a few changes, see last commit message. I'll run few test rounds...
@grahamwhaley , Thanks for your feedback! - Patches squashed into one. - `SMF_USE_COLLECTD=true` added and the string test changed to `true`. - Rmd script link fixed. Issue: - When I...