Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2024-01-09
16:25:17 kashyap Yeah
16:25:35 kashyap So it's about this specless blueprint - a suggestion that came from sean-k-mooney: https://blueprints.launchpad.net/nova/+spec/allow-disabling-ephemeral-disk-formatting
16:25:56 kashyap The main motivation (as I understand it) seems to be the following:
16:26:16 kashyap (1) To explore removal of libguestfs / and unneeded dependencies
16:26:38 kashyap (2) Doing so will help reduce security/package footprint from containers that might be shipping nova-compute.
16:27:20 kashyap Today libguestfs' usage seems to be in the area of file injection and ephemeral disk creation.
16:27:26 bauzas to be clear, the default would be "formatted", right?
16:27:42 kashyap I think so, yes. I need to check.
16:27:51 kashyap bauzas: That said, putting my upstream hat on:
16:28:25 bauzas we can't really change our current behaviour so I'd prefer to keep the current usage by default
16:28:56 bauzas but of course, if operators and products want to not use libguestfs, they can change the option value
16:29:19 kashyap We should also bear in mind that libguestfs provides many other "adjacent tools" (e.g. an incomplete list: https://paste.opendev.org/show/bfNahtW5QJVeYSQUNaFH/) that operators might be using.
16:29:35 bauzas there could be some nits on the values of the items for this config option
16:29:52 kashyap (These "adjacent tools" are all to do with disk images, etc. They're extremely useful. So even for downstreams I'd be wary of suggesting to 'drop' libguestfs.)
16:29:58 bauzas but looks to me some implementation question, not a design one
16:30:17 kashyap bauzas: Yeah, the nits could be worked out. But yeah, the devil is in the implementation indeed.
16:30:37 bauzas kashyap: do you remember the current defaults we have for ephemeral disks ? ext4 ?
16:31:12 kashyap Quickly checking...
16:31:19 bauzas this would mean that we would need to provide all the supported disk formats in the config option items
16:31:36 bauzas maybe people don't want this, right?
16:31:42 kashyap (I'd think so, it'd be one of ext3)
16:31:49 bauzas are we able to format our disks by something else but the default format ?
16:32:12 bauzas I can quickly look at the code
16:34:38 kashyap (There's several places the CONF.default_ephemeral_format is used - I guess you're looking for this?)
16:37:16 bauzas nope, I just wonder the negative value of 'unformatted' which then would be the default
16:38:10 kashyap bauzas: The current default is ext4, as I see it
16:38:23 bauzas everything seems to be handled by this module https://github.com/openstack/nova/blob/master/nova/virt/disk/vfs/guestfs.py
16:39:58 kashyap It depends on what you mean by "everything" :) I think you mean everything related to libguestfs. But we don't have to get into the details of it here.
16:39:59 bauzas anyway, I'm OK to approve this specless blueprint but we'll need to make sure that the default behaviour is continuing to support to format the ephemeral disk
16:40:15 bauzas sean-k-mooney: any concern by what I said ?
16:40:16 kashyap Please add that remark in the BP. I need to still explore what makes sense here
16:40:34 kashyap But also I need to hear if anyone actually needs this. We shouldn't do needless work
16:41:12 kashyap I have a hard stop in 5 minutes; afraid. I need to visit a friend in a hospital. Can discuss the rest afterwards
16:41:41 bauzas anyone not wanting to accept this blueprint ?
16:41:54 bauzas if not,
16:42:43 bauzas #agreed https://blueprints.launchpad.net/nova/+spec/allow-disabling-ephemeral-disk-formatting is approved as specless, provided it defaults to format the disks
16:42:49 bauzas kashyap: ^
16:43:03 kashyap Okay; for now we can go ahead; thank you
16:44:35 bauzas ++
16:44:42 bauzas auniyal: you wanted to discuss a specific bug ?
16:45:25 auniyal yeah - we got this bug https://bugs.launchpad.net/nova/+bug/2048154
16:45:42 auniyal similar to this - https://bugs.launchpad.net/nova/+bug/2048184
16:45:55 auniyal I have created a patch - https://review.opendev.org/c/openstack/nova/+/904817
16:46:24 bauzas add your patch proposals into https://etherpad.opendev.org/p/nova-caracal-status
16:46:41 auniyal sean-k-mooney for " source_type: snapshot and source_type:image"
16:46:47 auniyal is it snapshot image
16:47:28 auniyal I created snapshot of Vm then Vm from that snapshot, but their source_type was sa same as original vm
16:47:30 bauzas auniyal: add your patches into https://etherpad.opendev.org/p/nova-caracal-status#L114 so people could review them
16:47:34 auniyal ack bauzas
16:48:03 bauzas I think we're done for the meeting
16:48:15 bauzas auniyal: we don't really discuss bugs in the nova meeting
16:48:31 bauzas but if you want, please ping other folks after we end this meeting
16:48:55 auniyal ack
16:49:45 auniyal bauzas, I added the patch in etherpad
16:49:54 bauzas ok, thanks all then
16:50:02 bauzas and happy new year again
16:50:04 bauzas #endmeeting
16:50:04 opendevmeet Meeting ended Tue Jan 9 16:50:04 2024 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)
16:50:04 opendevmeet Minutes: https://meetings.opendev.org/meetings/nova/2024/nova.2024-01-09-16.00.html
16:50:04 opendevmeet Minutes (text): https://meetings.opendev.org/meetings/nova/2024/nova.2024-01-09-16.00.txt
16:50:04 opendevmeet Log: https://meetings.opendev.org/meetings/nova/2024/nova.2024-01-09-16.00.log.html
16:50:17 elodilles thanks o/
17:00:36 opendevreview Amit Uniyal proposed openstack/nova master: Refactor vf profile for PCI device https://review.opendev.org/c/openstack/nova/+/905134
17:01:28 sean-k-mooney auniyal: i was refering to a volume created form a volume snapshot
17:02:03 sean-k-mooney bauzas: sorry i was on another call
17:02:27 sean-k-mooney bauzas: but know i dont have an objection ot https://blueprints.launchpad.net/nova/+spec/allow-disabling-ephemeral-disk-formatting being specless
17:03:11 sean-k-mooney bauzas: and ya the default should not change from today
17:03:35 sean-k-mooney you shoudl just have the option of configuring it to be unformatted. and not require libguestfs as a result
17:03:41 sean-k-mooney but no change of behavior by defaul
17:05:16 opendevreview Amit Uniyal proposed openstack/nova stable/2023.2: Refactor vf profile for PCI device https://review.opendev.org/c/openstack/nova/+/905135
17:07:47 opendevreview Amit Uniyal proposed openstack/nova stable/2023.1: Refactor vf profile for PCI device https://review.opendev.org/c/openstack/nova/+/905137
17:08:33 opendevreview Amit Uniyal proposed openstack/nova stable/zed: Refactor vf profile for PCI device https://review.opendev.org/c/openstack/nova/+/905138
17:15:39 bauzas sean-k-mooney: my only implementation concern is what we would provide as a option value for continuing to format the disk
17:16:27 bauzas should it be "formatting" or "ext4" or something else
17:19:58 sean-k-mooney well the propsasl is to extend the existing enum value with unformated
17:20:09 sean-k-mooney and leave the existing behavior unchaged
17:20:20 sean-k-mooney the current formating is driver speciric
17:20:35 sean-k-mooney and we also look at the OS_TYPE metadta if set on the instnace image
17:21:06 sean-k-mooney for linux i belive we default to ext4 but use fat32 or ntfs for windows
17:21:14 sean-k-mooney non of that behavior needs to change
17:21:42 sean-k-mooney we just need to add unformated ot https://docs.openstack.org/nova/latest/configuration/config.html#DEFAULT.default_ephemeral_format
17:22:05 sean-k-mooney an yes so looking at the docs windows guest use ntfs by default
17:22:41 sean-k-mooney so the default will still be None meaning the virt driver chooses
17:23:21 sean-k-mooney bauzas: on a related note i want to remove some config options that should have been removed a long time ago but i need to deprecate them first.
17:24:02 sean-k-mooney obviously we need to agree to do that but is that something yoru happy to agree on via gerrirt or do you want me to file something
17:24:48 sean-k-mooney i want to for https://docs.openstack.org/nova/latest/configuration/config.html#DEFAULT.vif_plugging_is_fatal to be true and remove it as an option or at least move it to workarounds
17:25:25 sean-k-mooney you shoudl not run with vif_plugging_is_fatal=false in production
17:25:28 sean-k-mooney ever
17:25:58 sean-k-mooney so i want to deprecate that in this release and then condier removing it next cycle
17:26:42 opendevreview Amit Uniyal proposed openstack/nova master: Refactor vf profile for PCI device https://review.opendev.org/c/openstack/nova/+/905134
17:35:55 opendevreview Amit Uniyal proposed openstack/nova master: Refactor vf profile for PCI device https://review.opendev.org/c/openstack/nova/+/905134
18:40:36 opendevreview Merged openstack/nova stable/xena: Avoid n-cond startup abort for keystone failures https://review.opendev.org/c/openstack/nova/+/859000
19:04:04 sean-k-mooney bauzas: can you review this spec tomorrow its trivial but required as its an api change https://review.opendev.org/c/openstack/nova-specs/+/904368
20:38:02 opendevreview Merged openstack/nova stable/zed: Fix rebuild compute RPC API exception for rolling-upgrades https://review.opendev.org/c/openstack/nova/+/900341
#openstack-nova - 2024-01-10
02:26:54 opendevreview melanie witt proposed openstack/nova master: Enable use of native threads in scatter_gather_cells https://review.opendev.org/c/openstack/nova/+/650172
02:42:36 opendevreview melanie witt proposed openstack/nova master: libvirt: Introduce support for rbd with LUKS https://review.opendev.org/c/openstack/nova/+/889912
03:35:11 opendevreview melanie witt proposed openstack/nova master: Enable use of native threads in scatter_gather_cells https://review.opendev.org/c/openstack/nova/+/650172
05:00:59 opendevreview melanie witt proposed openstack/nova master: libvirt: Introduce support for raw with LUKS https://review.opendev.org/c/openstack/nova/+/884313
05:01:00 opendevreview melanie witt proposed openstack/nova master: libvirt: Introduce support for rbd with LUKS https://review.opendev.org/c/openstack/nova/+/889912
05:57:12 songwenping_ stephenfin: this patch https://review.opendev.org/c/openstack/openstacksdk/+/883238 pep8 check passed on my env,but online is always failed, pls help to review, tks.

Earlier   Later