| opendevreview | Abhishek Kekane proposed openstack/glance-specs master: Add spec to bypass chunked transfer of data https://review.opendev.org/c/openstack/glance-specs/+/947423 | 04:54 |
|---|---|---|
| opendevreview | Abhishek Kekane proposed openstack/glance master: Refactor scrubber to remove use of eventlet https://review.opendev.org/c/openstack/glance/+/947637 | 08:36 |
| opendevreview | Abhishek Kekane proposed openstack/glance master: Refactor scrubber to remove use of eventlet https://review.opendev.org/c/openstack/glance/+/947637 | 08:45 |
| -opendevstatus- NOTICE: notice Reminder: The Gerrit service on review.opendev.org will be offline for a server replacement maintenance between 16:00-17:00 UTC today per https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/message/D6VGKXHKXCV6TD6MFJY4H4KQBIM3AQYI/ | 14:14 | |
| opendevreview | Abhishek Kekane proposed openstack/glance-specs master: Add spec to bypass chunked transfer of data https://review.opendev.org/c/openstack/glance-specs/+/947423 | 14:15 |
| abhishekk | rosmaita: added you as a reviewer to a spec, we have ongoing discussion about size calculation at client side which needs to be answered, please have a look when you get some time | 14:45 |
| rosmaita | abhishekk: ack | 14:45 |
| abhishekk | thanks! | 14:45 |
| dansmith | abhishekk: rosmaita: yeah, maybe a video chat would be warranted here so we can get on the same page | 14:47 |
| rosmaita | i would not be averse to that | 14:48 |
| dansmith | free now? everyone is still on vacay it seems so I'm pretty free | 14:49 |
| dansmith | abhishekk: ^ | 14:49 |
| abhishekk | me not around but can join over mobile if we want to discuss it now | 14:50 |
| dansmith | you're not around? you seem around :) | 14:51 |
| abhishekk | I was, but left home few minutes before for some work :D | 14:51 |
| dansmith | ah okay | 14:51 |
| dansmith | either way, maybe rosmaita will have some important context for me to change my mind | 14:52 |
| rosmaita | do you want to talk about Scientology? | 14:52 |
| dansmith | um | 14:54 |
| dansmith | no? | 14:54 |
| rosmaita | ok, works for me ... want to chat now? | 14:54 |
| abhishekk | If you managed to discuss this then great, else we can revisit it tomorrow or during weekly meeting | 14:55 |
| dansmith | meet.google.com/knj-jbvv-etp | 14:55 |
| rosmaita | omw | 14:55 |
| dansmith | abhishekk: ^ if you want, but I can summarize on the review if there's anything useful | 14:55 |
| abhishekk | that will be great, | 14:56 |
| dansmith | abhishekk: rosmaita: summarized, check my work: https://review.opendev.org/c/openstack/glance-specs/+/947423 | 15:27 |
| rosmaita | ack | 15:27 |
| abhishekk | So are we going to provide another command line option for this? | 15:35 |
| dansmith | no option should be required.. what are you thinking would be necessary? | 15:36 |
| dansmith | if we're provided a filename to upload we just set the size on the image from that file | 15:36 |
| abhishekk | ack, will sparse image upload will have any impact here? | 15:40 |
| dansmith | how does that work? specifying a bunch of chunks with size and offset? | 15:41 |
| abhishekk | yes | 15:41 |
| dansmith | the client provides that? | 15:42 |
| dansmith | the spec I'm looking at makes it seem like glance just inspects the stream for holes and doesn't write them | 15:42 |
| dansmith | is this just the upload to the backing store from staging during import? | 15:44 |
| abhishekk | it is during upload as well | 15:45 |
| abhishekk | https://github.com/openstack/glance_store/blob/master/glance_store/_drivers/rbd.py#L616 | 15:45 |
| dansmith | yeah, so that's just it refusing to write chunks that are all zeroes right? | 15:45 |
| dansmith | not sure how this would change that at all | 15:45 |
| abhishekk | anyway we don't compare image_szie with bytes_written if image_size is provided | 15:46 |
| dansmith | that is already kindof a bug right? :) | 15:47 |
| abhishekk | :D | 15:48 |
| dansmith | looks to me like even if we upload with a content-length of 4 but we actually send 6, we'll just write the 6 and use that value instead of erroring about the inconsistency | 15:49 |
| dansmith | (on rbd, the others could be different I guess) | 15:49 |
| abhishekk | need to check others, I think file has similar code | 15:49 |
| dansmith | but that's broken right? | 15:50 |
| dansmith | like, we could probably say content-length=4 and then upload a constant stream of data until we fill the disk and DoS the store? | 15:51 |
| abhishekk | yes :/ | 15:53 |
| -opendevstatus- NOTICE: notice The Gerrit service on review.opendev.org is offline for a server replacement maintenance between 16:00-17:00 UTC today per https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/message/D6VGKXHKXCV6TD6MFJY4H4KQBIM3AQYI/ | 16:00 | |
| dansmith | abhishekk: https://bugs.launchpad.net/python-openstackclient/+bug/2107802 | 16:09 |
| abhishekk | ack | 16:10 |
| abhishekk | Seems like it does not happen if I provide --file /dev/zero so it must be something with the stdin handling. Just curious what happens with --file /dev/zero does image becomes active or rejects it due to format inspeection or before that? | 16:12 |
| dansmith | abhishekk: I just updated, I was wrong | 16:12 |
| abhishekk | ack | 16:12 |
| dansmith | I was trying glanceclient with --file not osc.. --file with osc behaves the same | 16:12 |
| abhishekk | ack | 16:13 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!