*** gyee has quit IRC | 00:14 | |
*** lixiaoy1 has joined #openstack-glance | 00:30 | |
*** Liang__ has joined #openstack-glance | 01:00 | |
*** bhagyashris has joined #openstack-glance | 01:08 | |
*** brinzhang has joined #openstack-glance | 01:13 | |
*** Liang__ is now known as LiangFang | 01:30 | |
*** mhen has quit IRC | 01:40 | |
*** mhen has joined #openstack-glance | 01:45 | |
*** MattMan_ has quit IRC | 01:47 | |
*** MattMan_ has joined #openstack-glance | 01:48 | |
*** itlinux has quit IRC | 03:17 | |
*** aniketh has joined #openstack-glance | 03:21 | |
*** lbragstad has quit IRC | 04:00 | |
*** udesale has joined #openstack-glance | 04:27 | |
*** brinzh has joined #openstack-glance | 04:33 | |
*** brinzhang has quit IRC | 04:36 | |
*** abhishekk has joined #openstack-glance | 04:43 | |
*** ratailor has joined #openstack-glance | 04:58 | |
*** kushalAgrawal has joined #openstack-glance | 05:31 | |
*** brinzhang has joined #openstack-glance | 05:43 | |
*** brinzh has quit IRC | 05:45 | |
*** pdeore has joined #openstack-glance | 05:56 | |
*** Luzi has joined #openstack-glance | 06:00 | |
*** rcernin has quit IRC | 07:00 | |
*** pcaruana has joined #openstack-glance | 07:04 | |
*** mosulica has joined #openstack-glance | 07:27 | |
*** ratailor_ has joined #openstack-glance | 07:56 | |
*** ratailor has quit IRC | 07:57 | |
*** lixiaoy1 has quit IRC | 08:54 | |
*** ratailor has joined #openstack-glance | 09:22 | |
*** ratailor_ has quit IRC | 09:23 | |
*** KeithMnemonic has joined #openstack-glance | 09:40 | |
*** LiangFang has quit IRC | 09:42 | |
*** ratailor_ has joined #openstack-glance | 09:43 | |
*** ratailor has quit IRC | 09:46 | |
*** bhagyashris has quit IRC | 10:04 | |
*** aniketh has quit IRC | 10:10 | |
*** KeithMnemonic has quit IRC | 10:16 | |
*** KeithMnemonic has joined #openstack-glance | 10:28 | |
*** udesale has quit IRC | 10:36 | |
*** KeithMnemonic has quit IRC | 10:42 | |
*** KeithMnemonic has joined #openstack-glance | 10:42 | |
*** KeithMnemonic has quit IRC | 11:01 | |
*** KeithMnemonic has joined #openstack-glance | 11:07 | |
*** abhishekk has quit IRC | 11:23 | |
*** aniketh has joined #openstack-glance | 11:40 | |
*** pdeore has quit IRC | 11:40 | |
*** openstackgerrit has joined #openstack-glance | 11:44 | |
openstackgerrit | wanghui proposed openstack/glance-specs master: Update tox minversion to 2.0 https://review.openstack.org/613284 | 11:44 |
---|---|---|
*** ratailor_ has quit IRC | 12:01 | |
*** jistr_ is now known as jistr | 12:14 | |
*** KeithMnemonic has quit IRC | 12:19 | |
*** zul has joined #openstack-glance | 12:19 | |
*** rosmaita has joined #openstack-glance | 12:25 | |
*** pcaruana has quit IRC | 12:26 | |
*** pcaruana has joined #openstack-glance | 12:39 | |
*** brinzhang has quit IRC | 12:53 | |
*** kushalAgrawal has quit IRC | 13:00 | |
*** pdeore has joined #openstack-glance | 13:14 | |
pdeore | Can I get some review on this spec-lite? https://review.openstack.org/#/c/609331/ please ? | 13:16 |
*** abhishekk has joined #openstack-glance | 13:17 | |
pdeore | jokke_, rosmaita, smcginnis ^^ | 13:17 |
smcginnis | pdeore: Looking... | 13:19 |
pdeore | smcginnis, Thanks!! | 13:19 |
rosmaita | pdeore: got a revision request, see the review | 13:20 |
pdeore | rosmaita, ack | 13:21 |
*** jistr is now known as jistr|call | 13:29 | |
*** d0ugal has quit IRC | 13:33 | |
*** lbragstad has joined #openstack-glance | 13:33 | |
*** d0ugal has joined #openstack-glance | 13:34 | |
*** pcaruana has quit IRC | 13:37 | |
*** Liang__ has joined #openstack-glance | 13:48 | |
*** d0ugal has quit IRC | 13:49 | |
*** Liang__ is now known as LiangFang | 13:49 | |
*** Luzi has quit IRC | 13:57 | |
*** rosmaita has quit IRC | 14:00 | |
*** rosmaita has joined #openstack-glance | 14:01 | |
*** pdeore has quit IRC | 14:01 | |
-openstackstatus- NOTICE: Zuul and Nodepool services are being restarted to migrate them to a new Zookeeper cluster. THis brings us an HA database running on newer servers. | 14:39 | |
*** jistr|call is now known as jistr | 14:50 | |
jokke_ | rosmaita: abhishekk smcginnis | 15:00 |
rosmaita | o/ | 15:01 |
abhishekk | o/ | 15:01 |
smcginnis | o/ | 15:01 |
jokke_ | ok ... so | 15:01 |
rosmaita | jokke_: smcginnis: need you two to approve https://review.openstack.org/#/c/612706/ | 15:01 |
*** Luzi has joined #openstack-glance | 15:01 | |
jokke_ | I think, like I wrote down to meeting channel, we need to backport, cut stable, then cut master | 15:02 |
rosmaita | jokke_: need you to approve https://review.openstack.org/#/c/613074/ | 15:02 |
rosmaita | those will finally turn off the periodic jobs in stable/rocky | 15:02 |
abhishekk | do I need to be around, if yes I will head back and will be back within 3 hours from now | 15:03 |
smcginnis | rosmaita: And the reason we weren't just removing those was just consistency that going forward we won't have to remember to do that every time a branch is cut? | 15:03 |
rosmaita | abhishekk: no, i think we will be ok | 15:03 |
rosmaita | smcginnis: correct | 15:03 |
smcginnis | abhishekk: Getting late there! | 15:03 |
abhishekk | rosmaita, smcginnis yes | 15:04 |
abhishekk | but I can be online, if required drop me a mail | 15:04 |
smcginnis | Thanks abhishekk | 15:04 |
jokke_ | thanks abhishekk get home safe | 15:05 |
abhishekk | yep, have a nice time ahead | 15:05 |
*** abhishekk has quit IRC | 15:05 | |
rosmaita | jokke_: smcginnis: backport of multihash-download-validation to stable/rocky glanceclient: https://review.openstack.org/#/c/613350/ | 15:05 |
rosmaita | smcginnis: do we need to do a glanceclient release from master for s-1? | 15:06 |
smcginnis | rosmaita: We can block that patch. Or we can update it to bump up the feature number to have the room in rocky. | 15:07 |
smcginnis | We will be generating release proposals every milestone for all cycle-with-intermediary libraries, so might be better to get it raised up and released now. | 15:08 |
rosmaita | ok, that's really what i was asking | 15:08 |
rosmaita | so: (1) we will backport to stable/rocky and release 2.13.0 from there; (2) we will update smcginnis patch to release 2.14.0 from master on nov 1 | 15:09 |
smcginnis | ++, good plan to me. | 15:11 |
jokke_ | sounds good to me | 15:11 |
rosmaita | smcginnis: so the glanceclient doesn't have to be released today, just before nov 1 ? | 15:11 |
jokke_ | smcginnis: glance Stein-1 release patch is up | 15:11 |
smcginnis | jokke_: Do we need that milestone release for openstack/glance for someone? | 15:11 |
smcginnis | Just making sure you saw http://lists.openstack.org/pipermail/openstack-dev/2018-September/135088.html | 15:11 |
jokke_ | oh I'm pretty sure I did not | 15:12 |
rosmaita | i definitely did not | 15:12 |
smcginnis | :) | 15:13 |
jokke_ | smcginnis: ohh | 15:13 |
rosmaita | so we do not have to do an S-1 today | 15:13 |
jokke_ | I'll abandone it | 15:13 |
jokke_ | rosmaita: ever | 15:13 |
smcginnis | Hopefully it means less work. | 15:13 |
rosmaita | it already has! | 15:13 |
smcginnis | ALways nice to have a deadline to keep things moving, but we hope to get rid of busy work if no one is actually consuming these milestone releases. | 15:14 |
jokke_ | rosmaita: what I'm reading from that mail is that the cycle-with-milestones release model is actually cycle-without-milestones | 15:14 |
rosmaita | :) | 15:14 |
smcginnis | We can still do them, but we shouldn't do them if it's just because someone arbitrarily said it's time to do so. | 15:14 |
rosmaita | i thnk smcginnis calls it "cycle-with-rc" | 15:14 |
smcginnis | Yep, I have a set of patches up to change the name. | 15:15 |
*** mosulica has quit IRC | 15:15 | |
jokke_ | I have mixed feelings about that | 15:17 |
jokke_ | but I wasn't there to argue about the decision | 15:17 |
jokke_ | this would have made sense 2+years ago when most of the release stuff was still manual work tbh :D | 15:17 |
rosmaita | smcginnis: did you have a release patch up for glance store? i thought i saw one | 15:20 |
smcginnis | rosmaita: Yes, let me find that. | 15:21 |
rosmaita | thanks! | 15:21 |
smcginnis | https://review.openstack.org/#/c/612838/ | 15:21 |
smcginnis | jokke_: Honestly, I wasn't entirely for it initially either. But I think once we started talking through it some more, I started liking it more and more. | 15:22 |
jokke_ | ok, I need coffee ... brb | 15:22 |
rosmaita | i thnk let's get abhishek's change to the config files in there before we release it | 15:22 |
smcginnis | jokke_: You can still respond with any concerns there. Nothings ever set in stone. | 15:22 |
smcginnis | rosmaita: OK, if you can -1 that and update it with the new hash, we can hold off. We have time. | 15:22 |
rosmaita | great, will do | 15:23 |
jokke_ | smcginnis: For me it actually helped last cycle to see what was done. When Launchpad helps you track stuff based on tags | 15:23 |
jokke_ | also I know there's companies who uses those for tracking/planning their internal work | 15:24 |
jokke_ | specially as doing release is literally 1min job to throw the patch up | 15:24 |
rosmaita | we need to make sure abhishek sees http://lists.openstack.org/pipermail/openstack-dev/2018-September/135088.html | 15:25 |
*** Luzi has quit IRC | 15:26 | |
rosmaita | just sent him an email | 15:26 |
-openstackstatus- NOTICE: The Zuul and Nodepool database transition is complete. Changes updated during the Zuul outage may need to be rechecked. | 15:30 | |
openstackgerrit | Liang Fang proposed openstack/glance-specs master: Spec Lite: Show Store Info https://review.openstack.org/612236 | 15:38 |
*** LiangFang has quit IRC | 15:46 | |
*** itlinux has joined #openstack-glance | 16:06 | |
*** gyee has joined #openstack-glance | 16:16 | |
*** jiaopengju has quit IRC | 16:18 | |
imacdonn | rosmaita: I just noticed a major error in my last rev of the spec-lite .. I'm not sure if you missed it, or read the error literally - I had "checksum will not be optional" ... I had changed it to "not required" then meant to change it to "optional", but neglected to remove the "not" :( | 16:19 |
*** jiaopengju has joined #openstack-glance | 16:21 | |
rosmaita | oh, i thought you said "not optional" on purpose | 16:24 |
rosmaita | i was wondering why you didn't just say "required", but decided not to nit pick | 16:24 |
rosmaita | so i read it as "all 3 are required" | 16:25 |
rosmaita | imacdonn: ^^ | 16:25 |
*** mnaser has quit IRC | 16:26 | |
rosmaita | i thought we agreed yesterday that all 3 would be required ... or did i stop paying attention at some point? | 16:26 |
imacdonn | rosmaita: OK, so we still need to reach agreement on that :/ | 16:26 |
*** mnaser has joined #openstack-glance | 16:26 | |
rosmaita | ok,just read through yesterday's discussion. looks like i agreed with myself that we should require all 3 | 16:29 |
imacdonn | I think it's horrible to force people to implement a legacy mechanism that will never be of any value (if cinder implements multihash in Stein) | 16:32 |
imacdonn | I think your argument is that images should look the same, whether created this way or via the "normal" methods | 16:32 |
imacdonn | if that's the case, we should be mandating validation_data ... otherwise we can have to some images with none at all | 16:33 |
rosmaita | we need to provide the checksum as a fallback in case the algo is not available to the client | 16:34 |
rosmaita | plus, anyone using older clouds will have tooling built around the checksum | 16:34 |
imacdonn | what sort of "client" do you have in mind here ? | 16:34 |
jokke_ | well that's why we stuck with the one configured algo | 16:35 |
jokke_ | so that the deployer knows what checksums it will get | 16:35 |
rosmaita | yes, but if the deployer picks a weird one, no guarantee the client will have access to it | 16:36 |
imacdonn | me interpretation was that multihash is based on hashlib, so any client is expected to either use hashlib or some equivalent that supports the same algorithms | 16:37 |
jokke_ | are you talking about client as in client used by nova and cinder or client as in client used by end user | 16:37 |
rosmaita | any consumer | 16:37 |
jokke_ | rosmaita: may I render you point invalid. One cannot set them at all if we mandate all 3 and the user doesn't have access to the algo the deployment is using | 16:38 |
rosmaita | that doesn't affect the consumer, though | 16:39 |
jokke_ | if deployer compiles openssl with different algos between glance and cinder/nova nodes, they can only blame themselves | 16:39 |
rosmaita | i dont' see how this is at all controversial. we have the md5sum as a fallback, let's use it | 16:39 |
jokke_ | and uses algo in glance that their conder/nova nodes doesn't support | 16:39 |
jokke_ | cinder | 16:39 |
imacdonn | we're conflating two separate issues here | 16:40 |
jokke_ | imacdonn: I do agree | 16:40 |
imacdonn | one is forcing a specific algo for multihash, and the other is whether or not checksum is forced/required | 16:40 |
imacdonn | personally, I hate it when software tells me whats best for me | 16:41 |
rosmaita | well, you will really hate this, because (1) the deployer picks the multihash algo, and (2) you must include md5sum as a backup for consumers who want to use it | 16:42 |
*** imacdonn has quit IRC | 16:42 | |
jokke_ | rosmaita: the 2 was never agreed when the multihash spec was agreed on | 16:43 |
rosmaita | guess he really hated it! | 16:43 |
rosmaita | jokke_: what do you mean? | 16:43 |
*** imacdonn has joined #openstack-glance | 16:43 | |
imacdonn | grr, stupid proxy server .. disconnects me every 24 hours :( | 16:43 |
imacdonn | anyway ... if this didn't make it through: I *did* include checksum, but I don't want to make it *required* | 16:43 |
rosmaita | in the spec: "We will leave MD5 calculations and checksums for backwards compatibility." | 16:44 |
jokke_ | rosmaita: the spec clearly states that the fallback for checksum is there for the cases multihash is not populated, not for "consumers who wants to use it" | 16:44 |
imacdonn | I wonder if config options could provide a way around this .... could add smth like acceptable_hashing_algorithms and checksum_required | 16:44 |
rosmaita | jokke_: i don't see that at all | 16:45 |
rosmaita | if that were the case, we wouldn't do md5 on new images | 16:45 |
imacdonn | you have to, though, if glance is calculating it ... either that, or you have to provide some way to make the decision about whether to calculate it or not | 16:46 |
imacdonn | based on either the request from the client, or maybe some config option (enable_checksum) | 16:46 |
imacdonn | without something like that, you have to assume that it might be needed... and since you have the image contents already, it's "cheap" to calculate it | 16:50 |
jokke_ | rosmaita: as said yesterday we shouldn't anymore after we get the consuming services upgraded to version that actually consumes the hashes. On the work items there is clear mention of "fallback to checksum field when not present", that's for those old images that never got the multihash populated. the only point when one might end up to problem where no verification is done is during upgrade if gl | 16:51 |
jokke_ | ance is upgraded first and nova+cinder is not running supported client | 16:51 |
imacdonn | whereas in the "locations" case, the consumer can make the decision about whether or not the legacy checksum is needed ... and they may NOT have the image content at-hand | 16:51 |
rosmaita | jokke_: read the problem description | 16:52 |
rosmaita | it mentions "broader" coverage, not replacement coverage | 16:52 |
rosmaita | also, "adding a new image will set all of checksum, os_hash_algo and os_hash_value fields" in the proposed change | 16:53 |
jokke_ | rosmaita: in that case we need to change the client to verify both | 16:53 |
jokke_ | because that's not what our implementaion does ;) | 16:53 |
jokke_ | as it stands in the current client implementation, the multihash replaced md5 operations with only fallback to md5 if multihash is not populated | 16:54 |
rosmaita | ok, so you can file a bug | 16:55 |
rosmaita | i still don't understand what the problem is with requiring all 3 | 16:55 |
rosmaita | jokke_: i think https://review.openstack.org/#/c/613350/ got caught in the zuul restart ... if i do 'recheck' will that be enough to get it picked up? | 16:57 |
rosmaita | at least i don't see it on http://zuul.openstack.org/status | 16:58 |
jokke_ | rosmaita: because the md5 sum is redundant and not provided anymore by majority of image providers. Requiring all three is literally just making users life difficult without any good reason so far as it's not used as soon as multihash is present and the deployment is running supported configuration of software (as of Stein release when this will be available) | 16:58 |
jokke_ | rosmaita: recheck should do the trick on those | 16:59 |
rosmaita | users -- who? someone who uses multiple openstack clouds and whose tooling is built around checksum? | 16:59 |
jokke_ | rosmaita: and that user can populate it in | 16:59 |
jokke_ | there is nothing in the proposal preventing providing the checksum | 17:00 |
imacdonn | the problem is that it places a pointless burden on ... someone .. either the image publisher or the end-user ... to download the image (some of our images are 50-80GB) and calculate a checksum that wil never actually be used for anything | 17:00 |
jokke_ | ^^ | 17:00 |
jokke_ | and for someone who is in panic "We need to get this updated image up there now!" that cycle is easily the difference of leaving the all the checksums out because that will just work | 17:01 |
jokke_ | which is promoting less secure behavior | 17:02 |
*** pcaruana has joined #openstack-glance | 17:03 | |
jokke_ | we do not have anything in the code that enforces any of the checksums being there. Cinder will create volume and Nova will try to boot intance without any complaints if none of the fields are populated | 17:05 |
rosmaita | yes, and that's too bad | 17:05 |
rosmaita | you can outvote me on the spec. i think for this particular spec, we should require all 3. | 17:07 |
openstackgerrit | iain MacDonnell proposed openstack/glance-specs master: spec-lite: Embed validation data in locations https://review.openstack.org/597648 | 17:13 |
openstackgerrit | Merged openstack/glance_store master: Provision to add new config options in sample config file https://review.openstack.org/603331 | 17:19 |
*** mvkr has quit IRC | 17:50 | |
*** zzzeek_ has joined #openstack-glance | 17:59 | |
*** itlinux has quit IRC | 18:52 | |
*** itlinux has joined #openstack-glance | 19:03 | |
*** aniketh has quit IRC | 19:38 | |
*** lbragstad has quit IRC | 19:43 | |
*** lbragstad has joined #openstack-glance | 19:43 | |
*** NostawRm has quit IRC | 19:45 | |
*** irclogbot_1 has joined #openstack-glance | 20:00 | |
*** irclogbot_1 has quit IRC | 20:22 | |
*** imacdonn has quit IRC | 20:31 | |
*** imacdonn has joined #openstack-glance | 20:32 | |
*** jcosmao has quit IRC | 20:35 | |
*** irclogbot_1 has joined #openstack-glance | 21:14 | |
*** itlinux has quit IRC | 21:52 | |
*** mvkr has joined #openstack-glance | 22:35 | |
*** brinzhang has joined #openstack-glance | 23:52 | |
*** gyee has quit IRC | 23:57 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!