opendevreview | Rajat Dhasmana proposed openstack/glance master: WIP: Test increasing per project quota to 50 https://review.opendev.org/c/openstack/glance/+/933550 | 09:08 |
---|---|---|
opendevreview | Rajat Dhasmana proposed openstack/glance master: WIP: Test increasing per project quota to 50 https://review.opendev.org/c/openstack/glance/+/933550 | 10:08 |
opendevreview | Markus Hentsch proposed openstack/glance master: Standardization of encrypted images https://review.opendev.org/c/openstack/glance/+/926295 | 14:20 |
opendevreview | Dan Smith proposed openstack/glance master: Make glance able to require disk_format match https://review.opendev.org/c/openstack/glance/+/930492 | 14:30 |
opendevreview | Dan Smith proposed openstack/glance master: Run safety checks on uploaded content https://review.opendev.org/c/openstack/glance/+/930646 | 14:30 |
abhishekk | dansmith: o/ | 14:49 |
dansmith | o/ | 14:49 |
abhishekk | regarding disk-format match where you are calling restore method to set image status to queued | 14:49 |
abhishekk | is there any chance that image data is stored on actual backend? | 14:50 |
abhishekk | because restore does not deletes it I guess | 14:50 |
dansmith | abhishekk: not sure what you mean.. during image data upload, I'm just aborting the transfer, like if we failed to write | 14:51 |
abhishekk | * partial image data I mean to say | 14:51 |
abhishekk | you are reading few chunks for this inspection right? | 14:51 |
dansmith | yes | 14:52 |
dansmith | the behavior should be the same as if we transfer a few chunks and then swift goes down, or something like that | 14:52 |
abhishekk | hmmm so stale data will be there or not, that is what I am suggesting | 14:53 |
abhishekk | May be I will configure file backend and verify | 14:53 |
dansmith | what does glance do if it reads a few chunks from the client and then the client dies? this should be the same behavior | 14:53 |
dansmith | I would expect you have to delete the image and re-create to get the state back, no? | 14:54 |
abhishekk | likely | 14:54 |
abhishekk | restore is not deleting it | 14:54 |
abhishekk | AFAIK if glance client dies in between upload the image remains in uploading state | 14:55 |
dansmith | sorry, is 'restore' a glance operation? | 14:57 |
abhishekk | it's a method which sets back image status to queued, https://review.opendev.org/c/openstack/glance/+/930492/9/glance/api/v2/image_data.py line #275 | 14:58 |
dansmith | okay, but then if you have to upload again it should overwrite? | 14:59 |
abhishekk | Hope so, never tried that | 14:59 |
* abhishekk will verify it | 15:01 | |
dansmith | could be a problem if not, independent of this inspection P) | 15:01 |
abhishekk | yes | 15:01 |
dansmith | abhishekk: unrelated, but I just noticed that disk_format is allowed to be None in the API schema.. what happens if you create an image as None and upload | 15:10 |
dansmith | ? | 15:10 |
abhishekk | It does not allow to upload | 15:11 |
dansmith | create but upload not allowed? | 15:12 |
dansmith | that's good, but.. why is that even allowed in the API? | 15:12 |
abhishekk | create but upload not allowed, right | 15:13 |
abhishekk | setting disk-format None? | 15:14 |
dansmith | abhishekk: yeah, why doesn't API require a value there if it's required before upload? | 15:15 |
abhishekk | I don't know the use case about that | 15:15 |
abhishekk | also upload call does not let you set the format so you need to use image-update patch call for it | 15:16 |
dansmith | yeah, I'm just curious why we let a user create with disk_format=null at all.. seems like a bad thing that could allow something to slip in | 15:17 |
abhishekk | hmm, may be rosmaita can shade some light on it | 15:19 |
dansmith | specifically this: https://github.com/openstack/glance/blob/master/glance/api/v2/images.py#L1920 | 15:21 |
abhishekk | Let me try to dig a little | 15:22 |
dansmith | it is not super important, I was just looking through where we enforce disk_format to be valid and noticed that we take CONF.disk_formats and explicitly add in None, which seems weird to me | 15:23 |
abhishekk | https://bugs.launchpad.net/glance/+bug/1398313 | 15:26 |
abhishekk | It does not have any reason mentioned there though | 15:27 |
abhishekk | may be in client it is an optional value and that is why this weird solution IMO | 15:29 |
dansmith | uh | 15:33 |
dansmith | that seems like a "fix the symptom not the problem" sort of thing | 15:33 |
abhishekk | :D | 15:33 |
opendevreview | Markus Hentsch proposed openstack/glance master: Standardization of encrypted images https://review.opendev.org/c/openstack/glance/+/926295 | 16:31 |
opendevreview | Dan Smith proposed openstack/glance master: WIP: Add GPT disk_format https://review.opendev.org/c/openstack/glance/+/933601 | 17:24 |
abhishekk | Currently glance gate is failing for glance-cinder-multistore-import job with quota exceeding issue. The quota issue seems to be occurring from a recent change in devstack (removing volumev3 endpoint) causing issues with openstack client commands (set quotas). The details of the issue and fix are in this release patch https://review.opendev.org/c/openstack/releases/+/933578 | 18:59 |
abhishekk | Until this release patch is merged, kindly avoid posting recheck on your patch. | 18:59 |
opendevreview | Rajat Dhasmana proposed openstack/glance master: Increase per project quota to 50 https://review.opendev.org/c/openstack/glance/+/933550 | 19:12 |
opendevreview | Rajat Dhasmana proposed openstack/glance master: Add new glance.wsgi module https://review.opendev.org/c/openstack/glance/+/932701 | 19:12 |
*** whoami-rajat_ is now known as whoami-rajat | 19:31 | |
opendevreview | Cyril Roelandt proposed openstack/glance master: Doc: update "Running Glance in HTTPD" section https://review.opendev.org/c/openstack/glance/+/933613 | 19:39 |
opendevreview | Cyril Roelandt proposed openstack/glance master: Do not test standalone deployments any more https://review.opendev.org/c/openstack/glance/+/933614 | 19:50 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!