| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-02-15 | |||
| 10:43:58 | lyarwood | kashyap: any idea why blkid wouldn't work within an instance using the q35 machine type? | |
| 10:46:17 | kashyap | lyarwood: It rings a faint bell; but my tired brain cannot seem to recall ... let me see if I can rediscover it | |
| 10:46:44 | sean-k-mooney | lyarwood: i have used it before | |
| 10:47:01 | lyarwood | kashyap: np, the context is https://review.opendev.org/c/openstack/nova/+/708701 where the only failures are due to the IDE bus not being present and these blkid failures to find the config drive | |
| 10:47:03 | sean-k-mooney | lyarwood: it normally works | |
| 10:47:10 | lyarwood | hmm ah wait I think I know what it is | |
| 10:47:21 | lyarwood | just rubber duck'd myself there | |
| 10:47:27 | lyarwood | the config drive is normally IDE | |
| 10:47:29 | sean-k-mooney | ide + q35 | |
| 10:47:36 | sean-k-mooney | ya its sata | |
| 10:47:36 | kashyap | lyarwood: Aaah | |
| 10:47:36 | lyarwood | really? | |
| 10:47:46 | sean-k-mooney | for q35 | |
| 10:47:49 | sean-k-mooney | so different name | |
| 10:47:54 | lyarwood | yeah so it's now SATA so blkid is going to be different | |
| 10:47:57 | lyarwood | yeah | |
| 10:47:58 | lyarwood | fun | |
| 10:48:00 | kashyap | lyarwood: Yes, for Q35 it is SATA -- you did the patch yourself :) | |
| 10:48:09 | lyarwood | yeah I knew the IDE thing | |
| 10:48:15 | lyarwood | I just didn't connect the dots with the config drive | |
| 10:48:39 | lyarwood | I thought the label would be the same tbh | |
| 10:48:45 | lyarwood | hmm | |
| 10:49:09 | sean-k-mooney | the filesystem lable shoudl be | |
| 11:20:30 | lyarwood | hmm can we kill the ovsdbapp.backend.ovs_idl.vlog DEBUG logs in devstack@n-cpu | |
| 11:20:49 | sean-k-mooney | ... yes i kind of ment to do that a while ago | |
| 11:21:15 | sean-k-mooney | its form the lib we use in os-vif | |
| 11:21:18 | lyarwood | ah | |
| 11:21:28 | sean-k-mooney | that said its not that annoying just a bit noisy | |
| 11:21:55 | sean-k-mooney | normally i dont notice it at this point but i perfonally have filtered it out in my head | |
| 11:22:15 | sean-k-mooney | we can register a log filter in the os-vif init | |
| 11:22:22 | sean-k-mooney | or do it in nova | |
| 11:22:59 | lyarwood | yeah it's just killing my devstack env and I assume isn't helpful for our log indexing | |
| 11:23:05 | sean-k-mooney | lyarwood: you are seeing this more recently because we now default to the ptyhon libvary | |
| 11:23:46 | sean-k-mooney | neutron should also have that log in teh l2 agent | |
| 11:24:40 | sean-k-mooney | if you feel like writing a patch it shoudl proably get setup here https://github.com/openstack/os-vif/blob/master/os_vif/__init__.py#L24-L49 | |
| 11:25:04 | lyarwood | yup I can look at that later today | |
| 11:27:36 | lyarwood | kashyap: any idea how I can tell if q35 has a SATA bus? | |
| 11:28:03 | kashyap | lyarwood: 'dmesg' in the guest? | |
| 11:28:11 | sean-k-mooney | well it will the way we generate the xml by default | |
| 11:28:48 | lyarwood | oh wait I wonder if support is actually missing in the cirros image | |
| 11:28:56 | lyarwood | everything looks fine in the domain | |
| 11:29:04 | lyarwood | the device is there and attached as a SATA cdrom etc | |
| 11:29:20 | lyarwood | it's just not there within the guestOS | |
| 11:29:21 | kashyap | lyarwood: Very good point; it's the CirrOS image | |
| 11:29:39 | sean-k-mooney | well the cirros image usess a ubuntu 18.04 kernel | |
| 11:29:44 | sean-k-mooney | so it will have sata support | |
| 11:29:49 | lyarwood | you would think | |
| 11:30:02 | kashyap | Yeah, don't assume. | |
| 11:30:02 | lyarwood | don't they recompile the kernel now? | |
| 11:30:07 | sean-k-mooney | im not | |
| 11:30:20 | sean-k-mooney | i have use it with sata before | |
| 11:31:00 | sean-k-mooney | although i dont know if i have used it with sata and q35 | |
| 11:31:17 | sean-k-mooney | i have used it with sata and pc | |
| 11:32:35 | sean-k-mooney | lyarwood: are you lookign at a gate failure by the way | |
| 11:33:13 | lyarwood | sean-k-mooney: yeah the tip of my q35 series switches nova-next over to using q35 and some tests that rely on config drive (device tagging etc) are failing | |
| 11:33:29 | sean-k-mooney | ah ok | |
| 11:33:42 | lyarwood | sean-k-mooney: have a local reproducer anyway and I can't see anything related to SATA in dmesg within the instance | |
| 11:37:19 | sean-k-mooney | ya its strang | |
| 11:37:23 | sean-k-mooney | i booted it too | |
| 11:37:29 | sean-k-mooney | so it boots fine with sata | |
| 11:37:36 | sean-k-mooney | but lsblk and blkid are empty | |
| 11:38:55 | lyarwood | f33 finds the device fine | |
| 11:39:07 | lyarwood | weird, I could make the cdrom bus host configurable? | |
| 11:39:13 | lyarwood | as a workaround | |
| 11:39:23 | sean-k-mooney | well its configurable via the image alredy | |
| 11:39:52 | sean-k-mooney | whats odd its ist not in /sys/block | |
| 11:40:50 | kashyap | [ 1.724456] ahci 0000:00:1f.2: AHCI 0001.0000 32 slots 6 ports 1.5 Gbps 0x3f impl SATA mode | |
| 11:40:50 | kashyap | [root@rawhide-1 ~]# dmesg | grep -i sata | head -1 | |
| 11:41:21 | sean-k-mooney | i dont see it in /dev either | |
| 11:41:54 | sean-k-mooney | kashyap empty | |
| 11:43:15 | lyarwood | I wonder if SCSI would be a better bus for this anyway | |
| 11:43:34 | kashyap | sean-k-mooney: Look for `dmesg | grep -i sata` | |
| 11:43:41 | lyarwood | ah I guess we don't know if we have the controller setup | |
| 11:43:41 | kashyap | What gues did you boot? | |
| 11:43:56 | sean-k-mooney | kashyap: i did and it gives nothing | |
| 11:44:06 | kashyap | What guest? | |
| 11:44:07 | lyarwood | yeah same for me with cirros | |
| 11:44:11 | sean-k-mooney | cirros | |
| 11:45:16 | kashyap | Right; my guest is Fedora ("Rawhide" - rolling) | |
| 11:45:51 | sean-k-mooney | yes which is not what we are trying to debug | |
| 11:46:12 | sean-k-mooney | its odd because im not seeing anyting in /dev | |
| 11:46:18 | sean-k-mooney | or /sys for the disk | |
| 11:46:36 | sean-k-mooney | im really confused how the root filestyem is currenlty mounted | |
| 11:47:02 | kashyap | sean-k-mooney: I was posting it for info; not saying Fedora is what we're debugging here | |
| 11:47:51 | kashyap | lyarwood: sean-k-mooney: I see some forums online say CirrOS does not support SATA interface | |
| 11:48:53 | sean-k-mooney | unless we use the vfat config driver format that woudl be a problem | |
| 11:49:33 | sean-k-mooney | we cant really assuem scsi will work and virtio-blk is not avalid for a cdrom | |
| 11:50:10 | lyarwood | sean-k-mooney: ah check /etc/modules | |
| 11:50:22 | sean-k-mooney | just deleted them to try scsi | |
| 11:50:45 | lyarwood | scsi should work but I think we can get sata to also work by updating /etc/modules | |
| 11:51:38 | sean-k-mooney | that woudl requrie cirros to be updated | |
| 11:52:03 | sean-k-mooney | given its botting it is able to use sata somehow | |
| 11:52:19 | sean-k-mooney | /etc/modules is just the set of addtional moduels to load | |
| 11:53:12 | sean-k-mooney | by the way why i said scsi isnt really an option is windows does not have teh virtio-scsi drivers by default | |
| 11:53:19 | sean-k-mooney | so we could not change the default to scsi | |
| 11:53:33 | sean-k-mooney | we chose sata becasue it shoudl work on all operating systmes | |
| 11:53:42 | lyarwood | is scsi == virtio-scsi? | |
| 11:53:51 | lyarwood | you have to set the model to virtio-scsi for that to be the case right? | |
| 11:54:01 | sean-k-mooney | yes you do | |
| 11:54:12 | sean-k-mooney | the other scsi contoller are all quite old | |