| *** chsc has joined #swift3 | 16:39 | |
| *** chsc has quit IRC | 16:39 | |
| *** chsc has joined #swift3 | 16:39 | |
| *** openstackgerrit has joined #swift3 | 21:01 | |
| openstackgerrit | Tim Burke proposed openstack/swift3 master: Enable H203 to prevent assertIs(None, ...) regressions https://review.openstack.org/455789 | 21:01 |
|---|---|---|
| kota_ | hello | 23:07 |
| kota_ | sorry for late a littl | 23:07 |
| kota_ | e | 23:07 |
| kota_ | timburke: ping | 23:09 |
| timburke | hi kota_! | 23:09 |
| timburke | thanks for the ping :-) | 23:09 |
| kota_ | :-) | 23:09 |
| kota_ | let's get started | 23:09 |
| kota_ | sorry, I didn't edit the wiki page and not so much for today | 23:10 |
| kota_ | but I'd like to hear any progress and want to catch up something to review | 23:10 |
| kota_ | I think I had some revies for the recent 2 weeks and almost of them landed... IIRC | 23:11 |
| kota_ | lemme check my dash board... | 23:11 |
| timburke | sure! i recently packaged a swiftstack release that pulled in some in-review patches; in my validation of that, i noticed a few of my own bugs in https://review.openstack.org/#/c/368264/ and https://review.openstack.org/#/c/448764/ | 23:12 |
| kota_ | oh, ok. | 23:13 |
| kota_ | both look nice to have (regarding to commit message) and look not so big | 23:14 |
| kota_ | will review soon as possible | 23:14 |
| timburke | and both had lurking bugs, so a more-thorough review will be nice :-) | 23:14 |
| timburke | thanks! | 23:14 |
| kota_ | timburke: if you think some patches with high priority (e.g. your package including them) please feel free to add me to the reviewers list | 23:15 |
| timburke | i also had an idea about how we could reduce our keystone requests by adding a STS-like endpoint (http://docs.aws.amazon.com/STS/latest/APIReference/API_GetSessionToken.html) which would do the normal authorization against keystone, then issue a token and secret that we could stuff in memcache to do subsequent authorizations | 23:15 |
| kota_ | i usually check my dashboard to find un-reviewed patch and adding me to the reviewer list will be helpful to notice. | 23:16 |
| timburke | i, er, ah... that's a lot of them :-) including somewhat embarrassingly (since it has no tests) https://review.openstack.org/#/c/411026/ | 23:17 |
| kota_ | sometimes i miss the notification for the new patch | 23:17 |
| timburke | sure, i'll go ahead and add you | 23:17 |
| kota_ | sorry the intruption, so get back to... | 23:17 |
| kota_ | STS-like endpoint? | 23:17 |
| kota_ | sounds like boto3 supports? | 23:18 |
| timburke | ...i think? i remember having it come up in a call with a prospective customer, that they were using that so applications could be issued temporary credentials from a central authority that knew longer-lived secrets | 23:20 |
| kota_ | timburke: i think that is useful because it helps the code not to require the secrets inside. | 23:21 |
| kota_ | just call for getting the secret credential and if the code has correct permission, it should work | 23:22 |
| kota_ | i think STS is sort of that, not sure. need to learn m@ore | 23:22 |
| kota_ | oh, no. my daughter touches my laptop!!! | 23:23 |
| timburke | hehe :-) | 23:23 |
| timburke | so i did something similar for our own auth system that integrates with ldap -- we can issue auth tokens and such by doing ldap queries that include the password, but we don't want to keep it in memory so subsequent requests won't still know it. so i let ldap users sign *using their currently-issued auth token*! | 23:24 |
| kota_ | hmm | 23:26 |
| timburke | other than that, i've seen some renewed interest in https://bugs.launchpad.net/swift3/+bug/1636663 and https://bugs.launchpad.net/swift3/+bug/1522578 from customers... | 23:28 |
| openstack | Launchpad bug 1636663 in Swift3 "A complete multipart upload request verification failed due to object-server unable to update container DB" [Undecided,New] | 23:28 |
| openstack | Launchpad bug 1522578 in Swift3 "Etag generation in swift3 does not match AWS S3" [Undecided,Confirmed] | 23:28 |
| timburke | i think that covers most of what i've been up to lately with swift3 :-) | 23:28 |
| kota_ | I think the latter one is known issue I tracked | 23:29 |
| kota_ | and the first one looks interesting | 23:29 |
| kota_ | ah, ic. request timeouts causes the error | 23:30 |
| timburke | yeah, the container servers get overloaded, i think, so some of the parts don't appear in the container listing yet | 23:31 |
| timburke | i think we can stop doing the listing now, though! we used to need it to get the content-length, but with recent-enough slo you can omit it | 23:31 |
| timburke | and we'll still have the etag from the client to pass through to verify the integrity of the slo | 23:32 |
| kota_ | gotcha, at least ocata release we have in the requirements, no length and etag needed to make a slo manifest. | 23:33 |
| timburke | i think we might just need to do one extra HEAD that we expect to 404 (for the object with the next upload ID) | 23:33 |
| kota_ | and even container not updated, the slo middleware will check the existence to the *object*-servers directly | 23:33 |
| timburke | yup :-) | 23:34 |
| kota_ | sounds reasonable | 23:34 |
| timburke | all done except for the typing :D | 23:35 |
| kota_ | ok. thanks | 23:35 |
| kota_ | ah, one thing. recently I've started to think to cut a release | 23:36 |
| timburke | good idea! if only for the security fix | 23:36 |
| kota_ | we have a bunch of changes since the last release. | 23:36 |
| kota_ | i think they we discussed today are not so rushed | 23:38 |
| kota_ | but I'll try to get them into the release | 23:38 |
| *** chsc has quit IRC | 23:38 | |
| kota_ | and prepare the change log in parallel | 23:39 |
| timburke | sounds good | 23:39 |
| kota_ | that will happen in the next weel or by the end of Apr | 23:39 |
| timburke | i'll keep an eye out for the patch | 23:39 |
| kota_ | thx | 23:39 |
| kota_ | that's all for today | 23:39 |
| timburke | thanks kota_! | 23:40 |
| kota_ | thanks timburke! | 23:40 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!