| *** sameo has joined #kata-dev | 05:14 | |
| *** dklyle has quit IRC | 06:35 | |
| *** sameo has quit IRC | 06:53 | |
| *** jodh has joined #kata-dev | 07:01 | |
| *** sgarzare has joined #kata-dev | 07:09 | |
| *** sameo has joined #kata-dev | 07:28 | |
| *** davidgiluk has joined #kata-dev | 08:05 | |
| *** gwhaley has joined #kata-dev | 08:08 | |
| *** pcaruana has quit IRC | 10:28 | |
| *** pcaruana has joined #kata-dev | 10:53 | |
| *** jugs has quit IRC | 11:20 | |
| *** jugs has joined #kata-dev | 11:26 | |
| *** hashar has joined #kata-dev | 11:52 | |
| *** devimc has joined #kata-dev | 12:27 | |
| *** vgoyal has joined #kata-dev | 12:41 | |
| *** pcaruana has quit IRC | 12:59 | |
| *** dklyle has joined #kata-dev | 13:00 | |
| *** pcaruana has joined #kata-dev | 13:02 | |
| *** devimc has quit IRC | 13:15 | |
| *** devimc has joined #kata-dev | 13:15 | |
| *** crobinso has joined #kata-dev | 13:42 | |
| *** hashar has quit IRC | 14:40 | |
| *** jodh has quit IRC | 14:59 | |
| *** sameo has quit IRC | 15:18 | |
| *** sgarzare has quit IRC | 16:53 | |
| kata-irc-bot | <fidencio> @archana.m.shinde https://github.com/kata-containers/tests/blob/master/integration/kubernetes/runtimeclass_workloads/kata-runtimeclass.yaml is the values we are using for the overhead on OpenShift | 16:58 |
|---|---|---|
| kata-irc-bot | <fidencio> Do you think we should, let's say, double it? | 16:58 |
| kata-irc-bot | <graham.whaley> @fidencio - ideally, measure it? @eric.ernst might be able to help guide how to do that? | 17:00 |
| kata-irc-bot | <fidencio> @eric.ernst, I'd be happy to learn how to properly measure how much we'd have to increase those values | 17:02 |
| kata-irc-bot | <fidencio> For a quick test, I'll double them and see if I get the same amount of failures with persistent volumes | 17:02 |
| kata-irc-bot | <archana.m.shinde> @fidencio Maybe @eric.ernst can provide more input on this, but just referring back to the setup I had used earlier, I had used a memory overhead of 256Mi | 17:11 |
| kata-irc-bot | <archana.m.shinde> so ya give that a shot and lets see how that goes | 17:12 |
| *** sameo has joined #kata-dev | 17:13 | |
| kata-irc-bot | <archana.m.shinde> @fidencio Wasnt around here yesterday, but do you have any failure logs that are seeing for the failing tests/ | 17:13 |
| kata-irc-bot | <archana.m.shinde> ? | 17:13 |
| kata-irc-bot | <fidencio> @archana.m.shinde: https://fidencio.fedorapeople.org/failures-kata-only.log | 17:16 |
| kata-irc-bot | <fidencio> and then the full logs of this specific run: https://fidencio.fedorapeople.org/202004251640/ | 17:16 |
| *** hashar has joined #kata-dev | 18:08 | |
| kata-irc-bot | <rlk> @archana.m.shinde wondering whether you have anything beyond https://github.com/kata-containers/tests/tree/master/metrics or any data stored I could take a look at? Thanks! | 19:20 |
| *** jugs has quit IRC | 19:22 | |
| kata-irc-bot | <graham.whaley> Hi @rlk. There are some non-kata-specific but k8s based metrics tests over at https://github.com/clearlinux/cloud-native-setup/tree/master/metrics that were used to look at scalability. I'm just working on updating the documentation there and then probably clean up the reporting tool as well, as there are some rough edges. | 19:24 |
| kata-irc-bot | <graham.whaley> We don't tend to publish or store actual data, as it is quite variable across systems and setups. We do use a few of the metrics you referenced in one of our CIs to try and catch regressions. Mostly for 'boot' times and memory size regressions | 19:25 |
| *** gwhaley has quit IRC | 19:30 | |
| *** davidgiluk has quit IRC | 19:31 | |
| *** jugs has joined #kata-dev | 19:33 | |
| fidencio | devimc: heya! | 19:47 |
| fidencio | devimc: so, my question is basically, do we have a "refcount not supported" or similar message when we try to use refcount on a system where we shouldn't? | 19:48 |
| fidencio | devimc: if so, I'd shamely check for that specific piece in the error message ... :-/ | 19:48 |
| *** jugs has quit IRC | 19:58 | |
| *** jugs has joined #kata-dev | 20:01 | |
| devimc | fidencio: got it, I don't remember the error message | 20:07 |
| fidencio | devimc: let me spawn a VM here and check for it | 20:10 |
| fidencio | Fedora should have XFS filesystems which are using reflink, IIRC | 20:10 |
| devimc | fidencio: I'm running a container | 20:12 |
| devimc | ;) | 20:12 |
| fidencio | devimc: is it enough? | 20:13 |
| devimc | fidencio: yes | 20:15 |
| devimc | unknown option -m reflink=0 | 20:17 |
| fidencio | devimc: cool, do you agree parsing for that an only performing the second part of the expr if the error is that one/ | 20:21 |
| devimc | fidencio: yeah - sounds good | 20:21 |
| fidencio | devimc: cool, thanks! | 20:21 |
| fidencio | devimc: and I know, it's ugly. I'm as ashamed as suggesting that as you'll be by writing that code. | 20:22 |
| fidencio | s/ as suggesting/ for suggesting/ | 20:22 |
| devimc | fidencio: yeh, it's ugly but there is no other way | 20:24 |
| devimc | fidencio: ptal | 20:28 |
| devimc | fidencio: should it be more specific ? | 20:29 |
| fidencio | devimc: looks good enough | 20:32 |
| devimc | fidencio: thanks | 20:32 |
| *** crobinso has quit IRC | 20:57 | |
| *** hashar has quit IRC | 21:01 | |
| *** devimc has quit IRC | 21:02 | |
| *** vgoyal has quit IRC | 21:40 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!