Thursday, 2018-09-27

tdasilvaDHE: what's your use case for swift? can you share?00:19
*** openstackgerrit has joined #openstack-swift00:19
openstackgerritMerged openstack/swift master: Configure diskfile per storage policy
*** two_tired2 has joined #openstack-swift00:51
*** mikecmpbll has quit IRC00:54
DHEtdasilva: we do HTTP-based video service.  to a certain degree I'm going to start serving content straight out of swift rather than having a big ZFS array feeding it...00:59
DHEthere will be some kind of proxy in front of it - 95% nginx - as it's not 100% static content, but it's damned close.01:00
DHEI'm sure I'll find other uses for swift once I have it, but that's what the main goal is...01:32
notmynameDHE: I'm guessing that's code for "porn"01:35
DHEwell it includes an IPTV Video On Demand library, so in the strictest sense there is porn on it...01:37
notmynameand it wouldn't be the first time swift has been used for that01:37
DHEwell, I haven't built it yet, but the porn is planned01:39
DHEyes I'm going to run the joke into the ground01:40
tdasilvaDHE: cool, where do you work?01:40
DHEISP in canada. includes a number of other over-copper services like TV, phone01:41
tdasilvayeah, neat! definetely a common use case01:46
openstackgerritMerged openstack/swift master: update 1space url in associated projects
openstackgerritMerged openstack/python-swiftclient master: fix tox python3 overrides
DHEI can see why. Letting swift take the brunt of the load works for me. :)02:02
openstackgerritTimur Alperovich proposed openstack/swift master: Handle non-ASCII characters in S3 v2 listings.
openstackgerritPete Zaitcev proposed openstack/swift master: Enable FakeConn to handle multiple requests/responses
*** psachin has joined #openstack-swift02:48
*** zaitcev has quit IRC02:59
*** rcernin_ has quit IRC03:42
*** rcernin has joined #openstack-swift03:43
*** two_tired2 has quit IRC03:54
*** hugokuo has joined #openstack-swift05:04
*** thurloat has quit IRC05:10
*** silor has joined #openstack-swift05:42
*** silor has quit IRC05:42
*** pcaruana has joined #openstack-swift06:33
*** rcernin has quit IRC07:12
openstackgerritSam Morrison proposed openstack/swift master: s3 secret caching
openstackgerritSam Morrison proposed openstack/swift master: s3api: Ensure secret is ascii in check_signature
*** gkadam has joined #openstack-swift07:36
*** mikecmpbll has joined #openstack-swift07:59
*** e0ne has joined #openstack-swift08:12
*** remixtj is now known as remix_tj08:27
openstackgerritMerged openstack/swift master: py3: port direct_client
openstackgerritMerged openstack/swift master: Replacing six.iter/dict.iter usages
*** gkadam is now known as gkadam-afk08:52
*** gkadam-afk is now known as gkadam08:59
*** mikecmpbll has quit IRC11:00
*** mikecmpbll has joined #openstack-swift11:01
*** pcaruana has quit IRC11:50
*** e0ne has quit IRC12:09
*** e0ne has joined #openstack-swift12:09
*** mikecmpbll has quit IRC12:12
*** mikecmpbll has joined #openstack-swift12:14
*** pcaruana has joined #openstack-swift12:39
*** e0ne has quit IRC13:22
*** e0ne has joined #openstack-swift13:26
*** psachin has quit IRC13:36
*** itlinux has quit IRC13:59
*** jistr is now known as jistr|call14:31
*** itlinux has joined #openstack-swift14:58
*** mikecmpbll has quit IRC15:00
*** mikecmpbll has joined #openstack-swift15:04
*** thurloat has joined #openstack-swift15:44
*** jistr|call is now known as jistr15:46
*** e0ne has quit IRC15:51
*** gyee has joined #openstack-swift15:53
*** gyee has quit IRC15:54
*** gyee has joined #openstack-swift15:57
*** hseipp has joined #openstack-swift16:10
*** hseipp has quit IRC16:20
*** e0ne has joined #openstack-swift16:52
*** e0ne has quit IRC16:54
notmynamegood morning17:02
*** mikecmpbll has quit IRC17:09
*** e0ne has joined #openstack-swift17:16
*** gkadam has quit IRC17:20
*** mvkr has quit IRC17:42
*** mikecmpbll has joined #openstack-swift18:08
timburkewe really ought to get some func tests going that exercise non-ascii accounts/users/passwords for tempauth... ideally that also cover s3api18:43
*** jistr has quit IRC18:47
*** jistr has joined #openstack-swift18:49
notmynametimburke: twitter thread this morning:
notmynameI wonder if there's any more debug info we could put around SignatureDoesNotMatch exceptions in s3request.py18:50
notmynamebut then again, I always struggle with that because you don't necessarily want to throw debug logging on every auth failure a client could do18:51
notmynamebut maybe we should do more in debug logs (there and other opaque failures)18:51
*** jistr has quit IRC19:08
*** jistr has joined #openstack-swift19:08
timburkenotmyname: i think that particular problem would've been helped by
patchbotpatch 575836 - swift - Give better errors for malformed credentials - 1 patch set19:09
timburkefwiw, i was just bumping into a region mismatch problem myself this morning...19:09
*** itlinux has quit IRC19:09
timburkehaving maybe probably helps, too (it certainly helped me), but it requires some carful inspection of the canonical request and string to sign19:11
patchbotpatch 575808 - swift - Add debugging info to SignatureDoesNotMatch responses (MERGED) - 5 patch sets19:11
*** jistr has quit IRC19:23
timburkeoh yeah, and since he'd mentioned swift3, he probably didn't even have that one...19:25
timburkethings get so much better when we actually merge patches :P19:25
*** jistr has joined #openstack-swift19:26
*** jistr has quit IRC19:28
*** jistr has joined #openstack-swift19:29
*** jistr has quit IRC19:34
*** jistr has joined #openstack-swift19:37
*** jistr has quit IRC19:39
*** e0ne has quit IRC19:42
*** e0ne has joined #openstack-swift19:47
*** jistr has joined #openstack-swift19:49
*** e0ne has quit IRC19:50
*** pcaruana has quit IRC19:54
openstackgerritTim Burke proposed openstack/swift master: Unquote URL before using splited parts.
*** e0ne has joined #openstack-swift20:08
openstackgerritJohn Dickinson proposed openstack/python-swiftclient master: py2 functional testing
notmynametimburke: simplest version of what we were talking about yesterday ^20:15
notmynameI want to see what the gate does with it20:15
timburkei think it won't be much ;-)20:22
notmynameugh (at myself)20:26
openstackgerritJohn Dickinson proposed openstack/python-swiftclient master: py2 functional testing
timburkeit's weird that we've got some periodic jobs listed on that've been queued for 60+ hours...20:49
notmynameperiodic stable too20:54
notmynameI'll let -infra know20:54
*** e0ne has quit IRC21:14
*** mvkr has joined #openstack-swift22:13
openstackgerritTimur Alperovich proposed openstack/swift master: Handle non-ASCII characters in S3 v2 listings.
*** rcernin has joined #openstack-swift22:29
notmynametimburke: sounds pretty scary22:36
openstackLaunchpad bug 1765679 in Swift3 "get_container_info may lose all sysmeta" [Undecided,In progress] - Assigned to Yuxin Wang (chhyx2008)22:36
notmynamedo you have a repro yet?22:37
notmynamefor "get_container_info will give you (and presumably cache!!) radically different results depending upon where in the pipeline it was called..."22:37
timburkenot yet, i can work on that. the really scary thing in my mind is that get_container_info *takes an app as an argument*22:38
timburkeso you can only trust it as far as you can trust the rest of the pipeline22:38
timburkei'll see what i can do about identifying a clear and demonstrable Bad arising from it -- i've got a suspicion that i'll be able to do something with ratelimit (which is early and does some get_container_info calls) and versioned_writes (which relies on sysmeta in a get_container_info response)22:45
timburkethe good news is that versioned_writes doesn't pop off the sysmeta -- so maybe you really *do* need s3api/swift3 to get in the bad state?22:46

Generated by 2.15.3 by Marius Gedminas - find it at!