| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-cyborg - 2020-03-19 | |||
| 06:02:30 | chenke | brinzhang_ | |
| 06:03:56 | brinzhang_ | chenke, maybe tomorrow? We can go to wechat ask Sundar ^^ | |
| 06:04:12 | brinzhang_ | he seems not in irc | |
| 06:04:28 | chenke | oh. I think he need to add you to core group | |
| 06:04:43 | brinzhang_ | chenke, thanks | |
| 06:05:01 | chenke | I remind him in wechat. | |
| 06:05:21 | brinzhang_ | chenke: yeah, thanks again :) | |
| 06:06:15 | chenke | np | |
| 06:42:06 | openstackgerrit | Merged openstack/cyborg master: Fix GPU's improper cpid_type allowed value and attach_handle_type value https://review.opendev.org/709433 | |
| 07:48:39 | openstackgerrit | Brin Zhang proposed openstack/cyborg master: Add Cyborg API Reference to Cyborg index page https://review.opendev.org/711869 | |
| 16:35:15 | Sundar | dansmith, sean-k-moone, gibi: I've been away for a few days. Starting today, I intend to make one final push to get the Cyborg series done. I have roughly a week to wrap things up in my current role and move on. It is good to see that every patch got some review. | |
| 16:35:33 | dansmith | Sundar: great thanks | |
| 16:37:13 | Sundar | dansmith: Thanks for your reviews. I am looking to include most of them. | |
| 16:37:43 | Sundar | Regarding your comment about not including a comment :): https://review.opendev.org/#/c/673735/45/nova/compute/api.py@2454 | |
| 16:38:35 | Sundar | first: ithe comment serves to explain what the term ARQ stands for, which will point the reader towards Cyborg. | |
| 16:39:17 | Sundar | second: I see that comments will less info have been approved :) https://review.opendev.org/gitweb?p=openstack/nova.git;f=nova/compute/api.py;hb=refs/changes/35/673735/45#l1193 | |
| 16:39:26 | Sundar | *s/will/with/ | |
| 16:39:47 | dansmith | Sundar: I figured you'd know by now, but that argument holds no weight with me | |
| 16:40:15 | Sundar | Even if you are the one that did the approving? :) | |
| 16:40:46 | dansmith | totally | |
| 16:40:59 | Sundar | Anyways, I think it is worth retaining some comments that expand the term ARQ, at least one in each major code path | |
| 16:46:21 | Sundar | Re. the IRC discussion http://eavesdrop.openstack.org/irclogs/%23openstack-nova/%23openstack-nova.2020-03-16.log.html#t2020-03-16T15:57:20, yes, I have tested the XML generated with real devices. Not only by looking at the xml, but also by running a sample application within the VM that interacts with the FPGA. | |
| 16:47:03 | Sundar | dansmith, sean-k-mooney: ^ | |
| 16:47:10 | dansmith | ack | |
| 16:48:09 | sean-k-mooney | Sundar: do you have any write up on how to do that | |
| 16:48:16 | sean-k-mooney | or where to get the sample code | |
| 16:48:53 | sean-k-mooney | there is a host that i should be able to get acess too if needed. | |
| 16:49:22 | sean-k-mooney | if i need to take over some of the nova work from you or someone else does knowing how to test it with real hardware would be useful | |
| 16:49:52 | sean-k-mooney | we have some openvino sampel applciation but i was hoping there was a simpelr hello world | |
| 16:54:21 | sean-k-mooney | Sundar: by the way does the nivida pgpu driver work. because i also have nvida gpus at home so i can also try testign with that. i dont have ones that support ther vgpu/nvida grid feature but this is just for standard physical gpu passhtough anyway | |
| 16:55:36 | sean-k-mooney | if i can test with real hardware while also not needing to deal with fpga programming that might be a win win since from a nova point of view it will be exactly the same | |
| 17:10:38 | Sundar | sean-k-mooney: The OPAE SDK comes with some sample bitstreams (they are under the hw/ dir, IIRC, which can be a bit confusing). They are the Native Loopback (nlb) apps, so they are named like nlb-0.gbs and nlb-3.gbs. There are corresponding testing apps named nlb0 and nlb3, On the host, the nlb-N.gbs needs to be programmed into the device, either | |
| 17:10:39 | Sundar | with Cyborg or by hand. The OPAE stack also needs to be installed in the VM. Then you can the testing apps like nlbN inside the VM. | |
| 17:11:40 | sean-k-mooney | Sundar: that is good to know | |
| 17:12:45 | Sundar | sean-k-mooney: I can send more detailed notes if needed. Re. GPUs, I think some Cyborg community folks from Lenovo tested one type of Nvidia GPU. They have not been active this cycle. | |
| 17:13:12 | Sundar | I'll ask others if they have tested with GPUs this cycle. | |
| 17:14:15 | sean-k-mooney | i need to reinsally my desktop that i use for dev which has some nvida gpus in it so i might give it a try. that my project for the weekend however | |
| 17:14:36 | sean-k-mooney | for somereason after the latest bios update it nolonger supprot booting form optane... | |
| 17:34:36 | dansmith | Sundar: to be clear, are you good through the end of next week or this week? | |
| 17:37:06 | Sundar | dansmith: through next week (till March 27). A few days here or there is fine. | |
| 17:37:19 | dansmith | gotcha | |
| #openstack-cyborg - 2020-03-20 | |||
| 10:48:23 | openstackgerrit | Merged openstack/cyborg master: Deployable V2 API implementation https://review.opendev.org/712835 | |
| #openstack-cyborg - 2020-03-21 | |||
| 00:53:55 | openstackgerrit | Merged openstack/cyborg master: Remove useless interfaces in agent https://review.opendev.org/710911 | |
| 01:02:51 | openstackgerrit | Merged openstack/cyborg master: support standard devstack plugin cloning https://review.opendev.org/708771 | |
| 01:10:20 | openstackgerrit | Merged openstack/cyborg master: move functions from devstack/plugin.sh to devstack/lib/cyborg https://review.opendev.org/709620 | |
| 01:10:21 | openstackgerrit | Merged openstack/cyborg master: simplify devstack plugin https://review.opendev.org/708773 | |
| 01:24:45 | openstackgerrit | Merged openstack/cyborg master: add support for installing cyborg client https://review.opendev.org/709629 | |
| 01:24:46 | openstackgerrit | Merged openstack/cyborg master: Fix setting cyborg agent hostname https://review.opendev.org/709749 | |
| 06:41:23 | openstackgerrit | Arthur Dayne proposed openstack/cyborg master: Generate PDF documentation https://review.opendev.org/686344 | |
| 07:53:44 | openstackgerrit | Brin Zhang proposed openstack/cyborg master: Generate PDF documentation https://review.opendev.org/686344 | |
| 07:53:45 | openstackgerrit | Brin Zhang proposed openstack/cyborg master: PDF documentation build depends packages https://review.opendev.org/714248 | |
| 08:15:00 | openstackgerrit | Brin Zhang proposed openstack/cyborg master: Generate PDF documentation https://review.opendev.org/686344 | |
| 14:44:51 | openstackgerrit | Brin Zhang proposed openstack/cyborg master: Generate PDF documentation https://review.opendev.org/686344 | |
| 15:03:04 | openstackgerrit | Brin Zhang proposed openstack/cyborg master: Generate PDF documentation https://review.opendev.org/686344 | |
| #openstack-cyborg - 2020-03-22 | |||
| 21:11:51 | openstackgerrit | Sundar Nadathur proposed openstack/cyborg master: Check during ARQ bind that the target instance has no other ARQs. https://review.opendev.org/712231 | |
| #openstack-cyborg - 2020-03-23 | |||
| 00:50:19 | openstackgerrit | Merged openstack/cyborg master: Generate PDF documentation https://review.opendev.org/686344 | |
| 15:05:32 | ildikov | Hi Cyborg Team | |
| 15:05:38 | ildikov | I have a quick question | |
| 15:06:00 | ildikov | From which release do you have the v2 API in place? | |
| 15:06:05 | ildikov | Is it Train or earlier? | |
| 18:42:14 | openstackgerrit | Merged openstack/cyborg-specs master: Change specs link and remove invalid setting https://review.opendev.org/711877 | |
| #openstack-cyborg - 2020-03-24 | |||
| 22:09:11 | openstackgerrit | Sundar Nadathur proposed openstack/cyborg master: WIP: Add UT and comments for deletion of ARQs by instance or ARQ UUIDs. https://review.opendev.org/714782 | |
| 23:11:19 | openstackgerrit | Sundar Nadathur proposed openstack/cyborg master: Check during ARQ bind that the target instance has no other ARQs. https://review.opendev.org/712231 | |
| #openstack-cyborg - 2020-03-25 | |||
| 15:27:46 | openstackgerrit | sean mooney proposed openstack/cyborg master: default host to socket.gethostname() https://review.opendev.org/714971 | |
| 15:29:05 | sean-k-mooney | dansmith: on your system is the output of "hostname" and "hostname --fqdn" different | |
| 15:29:37 | sean-k-mooney | now that your change has merged to remove the defaulting in the devstack plugin i need to have that ^ change | |
| 15:30:03 | dansmith | sean-k-mooney: does that match nova? | |
| 15:30:14 | sean-k-mooney | yes nova uses socket.gethostname | |
| 15:30:34 | sean-k-mooney | ill see if i can find the point in the code | |
| 15:30:37 | dansmith | sean-k-mooney: that system isn't booted up right now, so I'm not positive, but it probably have the default shortname in /etc/hosts for localhost, which might be the difference | |
| 15:32:29 | sean-k-mooney | dansmith: https://github.com/openstack/nova/blob/e80300ac20388890539a7f709e526a0a5ba8e63d/nova/conf/netconf.py#L52-L70 | |
| 15:33:41 | sean-k-mooney | if you have an fqdn in /etc/hostname that returns the fqdn i belive too | |
| 15:34:22 | sean-k-mooney | so with ooo the way it sets the fqdn in /etc/hostname means we get the fqdn back | |
| 15:34:45 | dansmith | mm, I don't think so, I think it does a dns query with the resolver and so if you have a fqdn first in /etc/hosts it'll return that | |
| 15:35:09 | sean-k-mooney | i guess i could try setting an fqdn and answer my own question | |
| 15:35:13 | sean-k-mooney | :) | |
| 15:40:36 | sean-k-mooney | so ya that works the way i expected http://paste.openstack.org/show/791144/ | |
| 15:43:02 | dansmith | hmm, okay, that's not what I would expect | |
| 15:43:13 | dansmith | ohh, sorry, | |
| 15:43:23 | dansmith | I was reading this wrong.. I thought we were doing gethostbyname() | |
| 15:43:24 | sean-k-mooney | well you are not ment to set the fqdn in /etc/hostname anyway | |
| 15:43:25 | dansmith | you're doing gethostname | |
| 15:43:45 | sean-k-mooney | yes we use gethostname not gethostbyname() | |
| 15:44:48 | sean-k-mooney | anyway no i have my multinode setup working again. | |
| 15:45:12 | sean-k-mooney | im going to test evacutate and try and test a rescudle with the fake dirver | |
| 15:45:31 | sean-k-mooney | i want to see if the allocation are handeld correctly | |
| #openstack-cyborg - 2020-03-26 | |||
| 03:00:34 | Sundar | Hello all | |
| 03:01:53 | s_shogo | Hi all | |
| 03:02:07 | Sundar | #startmeeting openstack-cyborg | |
| 03:02:08 | openstack | Meeting started Thu Mar 26 03:02:07 2020 UTC and is due to finish in 60 minutes. The chair is Sundar. Information about MeetBot at http://wiki.debian.org/MeetBot. | |
| 03:02:09 | openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | |
| 03:02:11 | openstack | The meeting name has been set to 'openstack_cyborg' | |
| 03:02:17 | Sundar | #topic Roll call | |
| 03:02:21 | Sundar | #info Sundar | |
| 03:02:31 | brinzhang_ | #brinzhang | |
| 03:02:40 | brinzhang_ | #info brinzhang_ | |
| 03:02:46 | s_shogo | #info s_shogo | |
| 03:03:27 | Sundar | Hi all, just the 3 of us? | |
| 03:03:53 | brinzhang_ | I think so ^ | |
| 03:04:11 | Sundar | Let's wait for a couple of minutes | |