| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2024-04-30 | |||
| 18:58:37 | sean-k-mooney | from where | |
| 18:58:39 | sean-k-mooney | https://paste.openstack.org/show/bcayBBbJaZjiagsjcVdu/ | |
| 18:58:55 | sean-k-mooney | im using a raw sting in that version to avoid any escaping | |
| 18:59:58 | sean-k-mooney | i guess you mean when this is used by nova | |
| 19:01:02 | frct1 | So somehow provider have to return escaped string, it is ? | |
| 19:01:52 | sean-k-mooney | the provider need to return a vaild json sring containing something that can be parsed by the client | |
| 19:02:28 | sean-k-mooney | cloud init is documeitnign that it shoudl have a srting with newlines excaped as \n | |
| 19:02:31 | sean-k-mooney | https://cloudinit.readthedocs.io/en/latest/reference/datasources/openstack.html#vendor-data | |
| 19:03:13 | sean-k-mooney | so the server litrally need to return "#cloud-config\npackage_upgrade: True\npackages:\n - htop" | |
| 19:03:21 | sean-k-mooney | including the " | |
| 19:04:26 | sean-k-mooney | my guess is internaly cloud-init is reading the content of that string directly adn parsing it in text mode | |
| 19:06:09 | sean-k-mooney | that or writign it to a file first and then reading it back | |
| 19:07:29 | sean-k-mooney | i think its doign this here ? https://github.com/canonical/cloud-init/blob/main/cloudinit/sources/DataSourceOpenStack.py#L193-L195 | |
| 19:08:21 | sean-k-mooney | so we are going to take this path https://github.com/canonical/cloud-init/blob/main/cloudinit/sources/__init__.py#L1154 | |
| 19:10:03 | sean-k-mooney | im not familar with this code but i think tis hten used here https://github.com/canonical/cloud-init/blob/7eb5a1df0769fb88bbe3fc357ab44a378770043b/cloudinit/stages.py#L562-L564 | |
| 19:11:08 | sean-k-mooney | which calls json_dumps on the string contaiing the cloud config | |
| 19:11:11 | sean-k-mooney | https://github.com/canonical/cloud-init/blob/7eb5a1df0769fb88bbe3fc357ab44a378770043b/cloudinit/stages.py#L604-L605 | |
| 19:11:45 | sean-k-mooney | and it finaly write it to a file | |
| 19:11:48 | sean-k-mooney | https://github.com/canonical/cloud-init/blob/7eb5a1df0769fb88bbe3fc357ab44a378770043b/cloudinit/stages.py#L590-L594 | |
| 19:12:14 | sean-k-mooney | which has mode wb https://github.com/canonical/cloud-init/blob/7eb5a1df0769fb88bbe3fc357ab44a378770043b/cloudinit/util.py#L2249 | |
| 19:12:37 | sean-k-mooney | so its doign a binary write of the raw data to a file | |
| 19:13:16 | sean-k-mooney | frct1: so the respocne form the backend woudl have to be a raw binary sring | |
| 19:13:31 | sean-k-mooney | contiing the cloud-init file content | |
| 19:14:14 | frct1 | Yep, i'm trying to implement it somehow | |
| 19:14:28 | frct1 | The main thing is to make vendor data provider return escaped data | |
| 19:28:41 | frct1 | Uhm, it works, lol https://paste.openstack.org/show/bShqKlZLG9hjmix2wCmc/ | |
| 19:29:55 | frct1 | https://paste.openstack.org/show/bV1JTGhzjKuRdsHGGS5B/, i will try to launch cloud-init shortly | |
| 19:33:24 | sean-k-mooney | so two thing. one your obviously in unsupproted terrirotry using implemenation specific knowldage | |
| 19:33:33 | sean-k-mooney | two if you get it work ing we have a few optiosn | |
| 19:33:48 | sean-k-mooney | first we could docuemnt and add test coverage for this hack | |
| 19:34:04 | sean-k-mooney | second we could still consider the new feature to do this properly | |
| 19:37:10 | frct1 | Actually i'm really inclined to add support for DynamicJSON provider's response typings like mentioned earlier "cloud-init:string@10.10.13.37:123,testing@10.10.13.37:124", so cloud-init key will be explicitly encoded as string, but testing will be presented as json | |
| 19:37:35 | frct1 | I will specs to provide details about observed hack to docs | |
| 19:37:45 | frct1 | I will add* | |
| 19:41:32 | frct1 | Wrapping everything up to update vendor data provider used with Consul and then move to compose docs update | |
| 19:42:55 | frct1 | What we have to add to test coverage for this case ? Something like check that string is probably valid escaped JSON string, i cant find any RFC that covers that | |
| 20:39:20 | melwitt | gmann: dunno if you may get a moment to take a look at this small API controllers tracking change https://review.opendev.org/c/openstack/nova/+/915732 I don't really understand it but I'm sure you will :) | |
| 20:46:06 | sean-k-mooney[m] | escaping is not really the correct way to thikn about it cloud-init treats the string as a binary sting | |
| 20:47:00 | sean-k-mooney[m] | python3 and json by defualt is generally parsed as utf8 | |
| 20:48:23 | sean-k-mooney[m] | what would likely be better then have cloud-init:string is perhaps cloud-init:base64 and then render the decoded content in | |
| 20:48:32 | sean-k-mooney[m] | the vendor data | |
| 20:52:30 | frct1 | You mean base64 for any service or especially for cloud-init ? What if other provider will return json encoded in base64 ? In what type we should decode it before response ? | |
| #openstack-nova - 2024-05-01 | |||
| 01:05:58 | opendevreview | Takashi Kajinami proposed openstack/python-novaclient master: Remove old excludes https://review.opendev.org/c/openstack/python-novaclient/+/917685 | |
| 05:36:50 | opendevreview | Rajesh Tailor proposed openstack/nova master: Fix KeyError on assisted snapshot call https://review.opendev.org/c/openstack/nova/+/900783 | |
| 11:55:58 | opendevreview | Merged openstack/python-novaclient master: Remove old excludes https://review.opendev.org/c/openstack/python-novaclient/+/917685 | |
| 12:33:57 | opendevreview | Merged openstack/placement master: Remove old excludes https://review.opendev.org/c/openstack/placement/+/917608 | |
| 14:48:36 | gmann | melwitt: sure, will check today | |
| 14:49:01 | gmann | I have opened that series but did not get chance to look into yet | |
| 15:09:17 | melwitt | gmann: ah gotcha. thanks :) | |
| 16:40:20 | opendevreview | Merged openstack/os-resource-classes master: Remove old excludes https://review.opendev.org/c/openstack/os-resource-classes/+/917614 | |
| 16:59:07 | opendevreview | Merged openstack/nova master: Remove old excludes https://review.opendev.org/c/openstack/nova/+/917594 | |
| #openstack-nova - 2024-05-02 | |||
| 10:53:29 | opendevreview | sean mooney proposed openstack/nova master: add bin dir for non eventlet console_scripts https://review.opendev.org/c/openstack/nova/+/904424 | |
| 10:53:29 | opendevreview | sean mooney proposed openstack/nova master: use futureist for scater_gather https://review.opendev.org/c/openstack/nova/+/905284 | |
| 10:53:30 | opendevreview | sean mooney proposed openstack/nova master: move context executor to utils and remove eventlet.tpool https://review.opendev.org/c/openstack/nova/+/905287 | |
| 10:57:55 | opendevreview | sean mooney proposed openstack/nova master: add bin dir for non eventlet console_scripts https://review.opendev.org/c/openstack/nova/+/904424 | |
| 10:57:56 | opendevreview | sean mooney proposed openstack/nova master: move context executor to utils and remove eventlet.tpool https://review.opendev.org/c/openstack/nova/+/905287 | |
| 10:57:56 | opendevreview | sean mooney proposed openstack/nova master: use futureist for scater_gather https://review.opendev.org/c/openstack/nova/+/905284 | |
| 11:19:06 | opendevreview | sean mooney proposed openstack/nova master: move context executor to utils and remove eventlet.tpool https://review.opendev.org/c/openstack/nova/+/905287 | |
| 11:45:32 | Luzi | Hi it would be good if someone from the Nova-Team (e.g. dansmith ) could look over the image encryption spec, thank you :) | |
| 13:26:16 | opendevreview | sean mooney proposed openstack/nova master: [WIP] use futerist for RBD calls https://review.opendev.org/c/openstack/nova/+/917962 | |
| 14:14:40 | opendevreview | Rajesh Tailor proposed openstack/nova master: Fix KeyError on assisted snapshot call https://review.opendev.org/c/openstack/nova/+/900783 | |
| 15:02:43 | opendevreview | Takashi Kajinami proposed openstack/nova master: doc: Add minimum QEMU/libvirt version for 2024.1 https://review.opendev.org/c/openstack/nova/+/917983 | |
| 16:38:58 | sean-k-mooney | dansmith: quick question. with the deivice alias stuff. i rememeber we checked the temest/service logs to confirm the aias was being used. do you know if we happen to have a tempet test that tries to use type=lun | |
| 16:39:19 | sean-k-mooney | if not it might be nice to add that to our test automation backlog | |
| 16:39:22 | dansmith | I'm sure not | |
| 16:39:30 | sean-k-mooney | ack | |
| 16:44:53 | sean-k-mooney | dansmith: as an fyi i rebased by eventlet stuff eiarler https://review.opendev.org/q/topic:%22eventlet-removal-part-1%22 so far so good. | |
| 16:55:07 | dansmith | sean-k-mooney: cool | |
| 16:56:11 | opendevreview | sean mooney proposed openstack/nova master: [WIP] use futerist for RBD calls https://review.opendev.org/c/openstack/nova/+/917962 | |
| 16:56:30 | sean-k-mooney | it might help if i actully ran the rbd commadn on the thread pool... | |
| 17:00:35 | opendevreview | sean mooney proposed openstack/nova master: [WIP] use futerist for RBD calls https://review.opendev.org/c/openstack/nova/+/917962 | |
| 17:16:43 | opendevreview | Balazs Gibizer proposed openstack/nova master: Reject AZ changes during aggregate add / remove host https://review.opendev.org/c/openstack/nova/+/821423 | |
| 18:39:17 | mnaser | sean-k-mooney: I know you brought up not long ago having CI for DPDK.. I'm a bit interested in that (even if we can 3rd party ci it?) .. is that something we can do over VMs or really need bm for this? | |
| #openstack-nova - 2024-05-03 | |||
| 00:52:48 | opendevreview | Merged openstack/nova master: api: Keep track of action controllers https://review.opendev.org/c/openstack/nova/+/915732 | |
| 09:32:42 | sean-k-mooney | mnaser: it can be done in vms. we technialy could do it upstream if we really wanted the limitation in the past was mianly around avialable memory in the zuul vms beucae both the vms and vswtich need to use hugepages which in the distant past need qemu to use kvm but that has not been required since 20.04 | |
| 09:33:08 | sean-k-mooney | now that dpdk is built into the distro ovs and we dont need to compile it form souce | |
| 09:33:24 | sean-k-mooney | all that is really requried is tweakign devstack to enable it when installing ovs | |
| 09:42:16 | mnaser | I see. I assume also the VMs need to be booted with huge pages enabled (the devstack “host” vm in this case too?) | |
| 13:03:55 | opendevreview | Rajesh Tailor proposed openstack/nova master: Handle neutron-client conflict https://review.opendev.org/c/openstack/nova/+/918048 | |
| 14:47:05 | spatel | Can I add hot-plug cpu in openstack nova? | |
| 14:47:07 | spatel | How does it work ? | |
| 14:47:09 | spatel | https://blueprints.launchpad.net/nova/+spec/nova-hot-add-vcpus-support | |
| 14:52:02 | greatgatsby | Hello. We're having problems with amphora migrations complaining about mtu 0 not matching mtu 9000? Haven't found much from googling, was hoping someone here might have some suggestions? | |
| 15:56:40 | supamatt | greatgatsby: why are you live migrating an amphora vm, just fail it over ;S | |
| 16:00:50 | opendevreview | Dan Smith proposed openstack/nova master: Avoid setting serial on raw LUN devices https://review.opendev.org/c/openstack/nova/+/918089 | |
| 16:01:03 | dansmith | sean-k-mooney: we also need this to enable the raw devices ^ | |
| 16:01:15 | dansmith | I guess that got lost in the focus on just the alias stuff | |
| 16:43:29 | dansmith | and just FYI, I've actually tested that with cinder and the LVM driver, and it seems to work. I get an actual scsi device in the guest that shows me the LIO target name and such | |
| 16:43:50 | dansmith | I thought that was going to require real hardware for some reason, but it seems to do the thing fine | |
| 18:03:38 | opendevreview | Dan Smith proposed openstack/nova master: Avoid setting serial on raw LUN devices https://review.opendev.org/c/openstack/nova/+/918089 | |
| #openstack-nova - 2024-05-04 | |||
| 10:31:54 | opendevreview | Kien Nguyen proposed openstack/python-novaclient master: support get and list server's metadata https://review.opendev.org/c/openstack/python-novaclient/+/915977 | |
| 14:10:08 | sean-k-mooney | dansmith: well the serial part was part of the followup of RMD/lun mode but ya i was not sure if we did that with the alias stuff but makes sense | |
| 19:26:56 | opendevreview | Vasyl Saienko proposed openstack/nova master: [ironic] Fix hard rebooting node https://review.opendev.org/c/openstack/nova/+/918195 | |
| 19:30:47 | opendevreview | Vasyl Saienko proposed openstack/nova master: [ironic] Fix hard rebooting node https://review.opendev.org/c/openstack/nova/+/918195 | |
| 19:34:27 | opendevreview | Vasyl Saienko proposed openstack/nova master: [ironic] Fix hard rebooting node https://review.opendev.org/c/openstack/nova/+/918195 | |
| 19:40:46 | opendevreview | Vasyl Saienko proposed openstack/nova master: [ironic] Fix rebooting node https://review.opendev.org/c/openstack/nova/+/918195 | |
| 19:45:26 | opendevreview | Vasyl Saienko proposed openstack/nova master: [ironic] Fix rebooting VM https://review.opendev.org/c/openstack/nova/+/918195 | |
| #openstack-nova - 2024-05-06 | |||
| 12:54:34 | greatgatsby | supamatt: this is during a failover. We're evacuating a node and failing over the amphora so it can be rebooted. Some amphora fail to failover to a new node with the mtu 0 vs 9000 error in the logs. Was just hoping someone might have seen this before. | |
| 15:42:42 | JayF | sean-k-mooney: https://review.opendev.org/c/openstack/nova/+/918195 would appreciate a review on that if/when you get a chance. Fixes a regression in Ironic driver from last cycle. I'm looking into what the impact would be on basic ops runtimes to add reboots to the mix. | |
| 17:13:02 | opendevreview | Sylvain Bauza proposed openstack/nova master: Add mtty/mdpy support for testing fake mdevs https://review.opendev.org/c/openstack/nova/+/898100 | |