Eero Tamminen

For starters, check for pause key could be at buffer swaps, and frame number could be show in the window title?

> Please help me address this issue. Only discrepancy I find in my container is the gid of the /dev/dri/renderD128 file is 'ssh' instead of 'render' in the host OS....

And a kernel supporting the GPU...

I'll need to update the docs a bit as upstream kernel has progressed further, but this ticket can be closed.

Tuomas fixed the deployment rule inconsistency in https://github.com/intel/intel-device-plugins-for-kubernetes/pull/1169 and Ukri is fixing overlay object names inconsistency in https://github.com/intel/intel-device-plugins-for-kubernetes/pull/1172. As to remaining namespace inconsistency... According to @ukri, it's better that manually...

Suggestions look fine to me. @mythi ?

What syscalls `collectd` is calling: `sudo strace -f -p $(pidof collectd)` ? And does `ltrace` show any library calls for it: `sudo ltrace -f -r --p $(pidof collectd)` ? PS....

What the FDs listed in those poll() calls are pointing to in: `ls -l /proc/$(pidof collectd)/fd/`? Kernel sysfs / devfs / procfs files? Network sockets? Something else? If you attach...

Ok, so polling is done by system curl library used by the `write_http` plugin. Looking at the `write_http.c` code, `curl_easy_perform()` is called only from `wh_post_nolock()`, which is just a wrapper...

clock_gettime() reminded me... Are any of your machines' clocks running faster than they should i.e. could they have been corrected backwards while collectd is running? Collectd seems to be using...