Monday, 2019-02-25

« Sunday, 2019-02-24 Index Tuesday, 2019-02-26 »
quantomworksyeah ive had no success01:05
quantomworksused branch 1.1301:05
quantomworksand k8s 1.13.301:05
quantomworksI have the vmx tag on both nodes01:05
*** igordc has quit IRC02:16
kata-irc-bot<krsna1729> @sfxworks the blog post is probably out of date03:50
kata-irc-bot<krsna1729> github link should work03:51
kata-irc-bot<krsna1729> can you check if crio is running?03:52
kata-irc-bot<krsna1729> if the deploy caused crio to fail then we need to file a bug03:53
*** stackedsax has joined #kata-dev03:56
*** fiddletwix has joined #kata-dev04:02
*** EricRen has joined #kata-dev06:59
*** jodh has joined #kata-dev07:51
*** sgarzare has joined #kata-dev07:53
*** quantomworks has quit IRC08:07
*** sameo has joined #kata-dev08:36
*** tmhoang has joined #kata-dev08:37
*** davidgiluk has joined #kata-dev09:00
*** gwhaley has joined #kata-dev09:07
kata-irc-bot<graham.whaley> hi @salvador.fuentes - it feels to me like the CIs are having a hard time - lots of fails. I'll have a look at the metrics machine, which I suspect maybe got its OBS repo URL re-injected badly from a stable branch build... and I'll probably PR the bare metal cleanup to try and avoid those in the future...11:58
*** sgarzare_ has joined #kata-dev12:03
*** sgarzare has quit IRC12:06
*** devimc has joined #kata-dev13:13
*** sgarzare__ has joined #kata-dev13:18
*** sgarzare_ has quit IRC13:21
*** sgarzare_ has joined #kata-dev13:36
*** sgarzare__ has quit IRC13:38
*** sgarzare_ has quit IRC13:38
*** sgarzare has joined #kata-dev13:41
kata-irc-bot<salvador.fuentes> hi @graham.whaley thanks, yeah, also I've seen some k8s tests failing, which I'm taking a look, though hard to reproduce locally14:28
kata-irc-bot<salvador.fuentes> @mnaser hi, we are experiencing issues connecting to the jenkins server, do you know if there is any issue on your side?15:42
kata-irc-bot<salvador.fuentes> seems like jenkins is back15:52
*** tmhoang has quit IRC16:03
kata-irc-bot<mnaser> yeah not sure whats there, ill have to dig a bit more16:29
*** dklyle_ has quit IRC16:40
*** dklyle has joined #kata-dev16:55
kata-irc-bot<sebastien.boeuf> stefanha dukgilbert I am trying latest virtio-fs code (both latest versions of qemu/kernel/virtiofsd), but it does not work, I get a QEMU error: ``` kvm_mem_ioeventfd_add: error adding ioeventfd: No space left on device ```17:06
kata-irc-bot<sebastien.boeuf> not sure what I'm doing wrong here. Or is it something you're aware about?17:06
kata-irc-bot<sebastien.boeuf> dgiluk ^17:09
kata-irc-bot<sebastien.boeuf> @dgiluk @stefanha actually this does not look specific to the virtio-fs work. Here is a very recent issue opened by Paolo: https://www.mail-archive.com/qemu-devel@nongnu.org/msg591919.html17:13
kata-irc-bot<sebastien.boeuf> let me know if you have any workaround for this cause I'd like to make sure I'm testing with your latest code17:14
kata-irc-bot<sebastien.boeuf> nm, I found the 2 lines patch on this same thread17:19
*** sgarzare_ has joined #kata-dev17:21
*** sameo has quit IRC17:23
*** sgarzare has quit IRC17:24
gwhaleydavidgiluk ^^17:35
gwhaley(drdavid sits on irc, looks like the slack<->irc bot was not translating names well....)17:36
davidgilukoh that bug...17:36
davidgiluksebastien.boeuf: Sorry, that's an upstream qemu bug17:37
kata-irc-bot<sebastien.boeuf> hey davidgiluk! yeah I just figured that...17:37
kata-irc-bot<sebastien.boeuf> that's alright, I patched my version locally17:37
davidgilukOK, cool; I do have that fix sitting on my tree here17:37
davidgilukit was working fine for me with virtio-fs until I reboot the guest17:37
kata-irc-bot<sebastien.boeuf> but now that you're here, I have question about virtio-fs17:38
kata-irc-bot<sebastien.boeuf> I cannot get the DAX use case to work17:38
kata-irc-bot<sebastien.boeuf> I mean it seems like it falls back to the regular use case all the time17:38
kata-irc-bot<sebastien.boeuf> I can't see the lo_setupmapping happening17:38
kata-irc-bot<sebastien.boeuf> only some lo_read()17:39
davidgilukyeh, now I guess you've passed the 'dax' flag to mount?17:40
kata-irc-bot<sebastien.boeuf> yes I did17:40
davidgilukand on the qemu command line on the -device vhost-user-fs-pci you added a ,cache-size=2G (give or take?)17:40
kata-irc-bot<sebastien.boeuf> 512M but yes I did17:41
davidgilukok, so the next bet is one of the zillion kernel config options17:41
kata-irc-bot<sebastien.boeuf> yeah that was my guess because my understanding here is that the driver does not use the DAX caps here17:42
kata-irc-bot<sebastien.boeuf> it still falls back onto the default usage17:42
davidgilukyeh, now I remember it did take me a while to get those right, let me just see if I can find that17:42
kata-irc-bot<sebastien.boeuf> thanks!17:43
davidgiluksebastien.boeuf: Have you got CONFIG_DAX_DRIVER, and CONFIG_FS_DAX set?17:43
kata-irc-bot<sebastien.boeuf> on the official doc, you mentioned about CONFIG_VIRTIO CONFIG_VIRTIO_FS CONFIG_DAX CONFIG_FS_DAX CONFIG_DAX_DRIVER CONFIG_ZONE_DEVICE17:43
kata-irc-bot<sebastien.boeuf> so yes I have all those enabled17:43
davidgilukthe others I have are FS_DAX_PMD and NVDIMM_DAX - but I can't remember if those are actually required17:44
kata-irc-bot<sebastien.boeuf> oh wait, do you have TRANSPARENT_HUGEPAGE ?17:45
davidgilukI do - but I'm not expecting that to be relevant17:45
davidgilukCan you try executing a binary from the mount rather than just reading/writing from a file (i.e. try an mmap)17:46
kata-irc-bot<sebastien.boeuf> okay I can try that17:46
kata-irc-bot<sebastien.boeuf> rebuilding with a few extra configs right now, I'll try the mmap asap17:48
davidgilukchrooting into a mount normally is the way I check17:54
kata-irc-bot<sebastien.boeuf> davidgiluk it works. Doing an mmap of a file part of the mount, it triggers lo_setupmapping18:00
kata-irc-bot<sebastien.boeuf> thanks ;)18:00
kata-irc-bot<sebastien.boeuf> I thought it would also be the case when we want to read from files directly18:01
kata-irc-bot<sebastien.boeuf> but I guess this means you would need to do one more tweak into the virtio-fs driver18:01
kata-irc-bot<sebastien.boeuf> what's the reason why you left regular file read out the lo_setupmapping scope?18:02
davidgilukAh ok, so I think some of the different caching modes will do it during a read18:02
davidgilukI'd need to check with vgoyal exactly which cache= is needed to do normal reads18:02
*** jodh has quit IRC18:03
kata-irc-bot<sebastien.boeuf> oh yeah that makes sense, I am using cache=none FWIW18:04
kata-irc-bot<sebastien.boeuf> okay anyway, thanks for your help davidgiluk, I can move forward with experimenting this for Firecracker now18:08
*** sgarzare_ has quit IRC18:08
davidgiluksebastien.boeuf: vgoyal says cache=always should do all files18:10
*** gwhaley has quit IRC18:12
kata-irc-bot<sebastien.boeuf> nice, let me try18:13
kata-irc-bot<sebastien.boeuf> yes I confirm this triggers the lo_setupmapping in this case18:14
kata-irc-bot<sebastien.boeuf> even for a simple file read18:14
davidgilukgreat18:16
*** tmhoang has joined #kata-dev18:28
*** devimc has quit IRC18:36
*** devimc has joined #kata-dev19:07
*** igordc has joined #kata-dev19:07
*** eguan has quit IRC20:09
*** eguan has joined #kata-dev20:11
*** davidgiluk has quit IRC20:11
*** sameo has joined #kata-dev21:19
kata-irc-bot<eric.ernst> @sfxworks around?  Did you open issue with your failure?  I'm curious what's failing.  Does CRIO come back up after running?21:47
kata-irc-bot<eric.ernst> @jose.carlos.venegas.m ping?22:00
kata-irc-bot<jose.carlos.venegas.m> @eric.ernst pong22:01
kata-irc-bot<eric.ernst> per @krsna1729 suggestion, i was thinking of adding a bot to packaging to manage our ToC for .mds22:01
kata-irc-bot<eric.ernst> at least for kata-deploy.22:01
kata-irc-bot<eric.ernst> Since I can't manage to keep it up to date.22:01
kata-irc-bot<eric.ernst> WDYT?  https://github.com/pholleran/toc-me22:01
kata-irc-bot<eric.ernst> Ah, but I guess it'll actually try to do a commit.22:02
kata-irc-bot<jose.carlos.venegas.m> me looks22:02
kata-irc-bot<eric.ernst> not sure if our process would appreciate this.22:02
kata-irc-bot<jose.carlos.venegas.m> It' be better if could rice a PR first22:03
kata-irc-bot<jose.carlos.venegas.m> I think no?22:03
kata-irc-bot<jose.carlos.venegas.m> but +1 automation22:04
kata-irc-bot<eric.ernst> Yeah...22:04
kata-irc-bot<jose.carlos.venegas.m> sorry I dont get the issue, so you want to update the toc from kata-deploy (in packaging automatically?)22:05
kata-irc-bot<krsna1729> i think across all repos having the toc update automatically is the issue22:06
kata-irc-bot<krsna1729> his update to kata-deploy md borke the toc22:06
kata-irc-bot<krsna1729> which is why we were looking at automation22:06
kata-irc-bot<jose.carlos.venegas.m> oh for any sounds good, so I imagine kata-deploy first as PoC22:09
kata-irc-bot<krsna1729> correct22:09
kata-irc-bot<jose.carlos.venegas.m> this is an issue I hit and james checks all the time22:09
kata-irc-bot<jose.carlos.venegas.m> lets ask James this may do its life easier22:09
kata-irc-bot<krsna1729> @james.o.hunt22:09
kata-irc-bot<jose.carlos.venegas.m> @krsna1729 I think he is offline now, need to wait until tomorrow22:12
kata-irc-bot<krsna1729> yup just tagging :slightly_smiling_face:22:13
*** sameo has quit IRC22:18
*** jdewald has joined #kata-dev22:20
kata-irc-bot<eric.ernst> @jose.carlos.venegas.m ptal: https://github.com/kata-containers/packaging/pull/36322:24
kata-irc-bot<jose.carlos.venegas.m> looks22:25
kata-irc-bot<jose.carlos.venegas.m> now I see the value on the bot :P22:26
kata-irc-bot<eric.ernst> :slightly_smiling_face:22:26
kata-irc-bot<eric.ernst> i'm nnot as good as a dumb bot22:26
kata-irc-bot<jose.carlos.venegas.m> they are going to take the world :anguished:22:27
kata-irc-bot<jose.carlos.venegas.m> @salvador.fuentes good eye on the typos22:30
*** devimc has quit IRC22:56
*** tmhoang has quit IRC23:29
« Sunday, 2019-02-24 Index Tuesday, 2019-02-26 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!