You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First off- cool project. You make it easy to benchmark. Thank you! Planning to continue digging in but wanted to relay a pain point during initial exploration.
In my usage I'm finding the default fio job does not produce latency numbers. In contrast, I notice latency is included when running axboe/fio/examples/latency-profile.fio. I start to think this kubestr behavior is expected and there's likely FIO complexities (axboe/fio/HOWTO.rst) which I haven't yet grasped.
Answering my own question, global option gtod_reduce=1 leads to the absence of latency stats. It's not yet clear what the drawback would be excluding this option.
First off- cool project. You make it easy to benchmark. Thank you! Planning to continue digging in but wanted to relay a pain point during initial exploration.
In my usage I'm finding the default fio job does not produce latency numbers. In contrast, I notice latency is included when running axboe/fio/examples/latency-profile.fio. I start to think this kubestr behavior is expected and there's likely FIO complexities (axboe/fio/HOWTO.rst) which I haven't yet grasped.
kubestr.default.json.txt attached, created with kubestr v0.4.36 against Azure AKS using:
As a cluster admin, seeing latency from the default job would grant meaningful insight.
The text was updated successfully, but these errors were encountered: