Robert Krawitz
Robert Krawitz
Might be interesting to add to clusterbuster...
OK, I've added code to ClusterBuster to do this (except for threading). ``` clusterbuster -T pod -P log options... ``` The options you care about are ``` --log-bytes-per-line= Number of...
Try that last again... ``` clusterbuster -T pod -P log --log-bytes-per-line=100 --log-bytes-per-io=1000000 --log-xfer-count=5 --log-processes=10 ```
This issue is plaguing me trying to start mssql-server inside a container inside a VM. The VM has 17 cores allocated to it, but /proc/cpuinfo has this: ``` sh-5.1# grep...
It appears that the exact contents of /proc/cpuinfo varies from run to run. Same thing happens if I use a VM with 16 cores. Once in a while I get...
Upon closer inspection, it turns out that the runtime that we're using is offlining the processors that are not showing up (as seen by lscpu). So for example, from a...
Actually, the real cluster is running 1.1; 1.0.2 it works fine.
I am not familiar enough with the innards of kube-burner to review in a lot of detail, but the approach certainly looks like the right one, so /lgtm other than...
I have a bit more informatoin on this. It appears that it was triggered when the backend mssql database failed. I copied strace into the workload pod and found this:...