| *** threestrands has joined #openstack-swift | 00:01 | |
| *** manous_ has quit IRC | 00:03 | |
| *** itlinux has joined #openstack-swift | 00:03 | |
| *** mat128 has quit IRC | 00:04 | |
| kota_ | clayg: AFAIK, that is from your suggestion that you were working there with tdasilva... | 00:13 |
|---|---|---|
| kota_ | To redirect, that's nice to have non conditional response always for Symlink object, AFAIK | 00:15 |
| *** openstackgerrit has joined #openstack-swift | 00:17 | |
| *** manous_ has joined #openstack-swift | 00:17 | |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: Add probe test to delete then revive a sharded container https://review.openstack.org/526552 | 00:17 |
| kota_ | Oh you mind on proxy controllers obj. I should be back to there. | 00:21 |
| kota_ | One possible point is because ec case... but anyway i need my laptop to check my thought. | 00:23 |
| kota_ | Mobile is too hard to review the code. | 00:24 |
| *** rcernin has quit IRC | 00:25 | |
| *** rcernin has joined #openstack-swift | 00:25 | |
| clayg | yeah so the 412 response *already* includes the X-Object-Sysmeta-Symlink-Target | 00:29 |
| clayg | kota_: the behavior is correct, but the implementation has some unfortunate side effects - luckily it looks like it will be easy to clean up - may not be a blocker! | 00:30 |
| timburke | yup. so mware can do whatever it wants with that -- try again at the new location, 3xx with a Location header, w/e | 00:30 |
| kota_ | clayg: sounds nice. Thanks! | 00:31 |
| timburke | oh... i was thinking that happened in https://review.openstack.org/#/c/357559/ -- must've been thinking of another change | 00:32 |
| patchbot | patch 357559 - swift - Include object sysmeta in POST responses (MERGED) | 00:32 |
| timburke | i know that we *needed* that for symlinks... so we could 307... maybe we were already sending back the meta? always had been? *shrug* | 00:33 |
| kota_ | clayg, timburk: if you leave anything more you find on Symlink, i will try to address them today (in my day time) | 00:33 |
| *** manous_ has quit IRC | 00:35 | |
| timburke | do we... include user meta in 416/412/304 responses? | 00:35 |
| clayg | timburke: we kinda shouldn't... | 00:36 |
| clayg | swift cli does not support upload -m | 00:38 |
| clayg | timburke: we DO include user-meta in 412 response | 00:39 |
| timburke | https://review.openstack.org/#/c/481202/ | 00:39 |
| patchbot | patch 481202 - python-swiftclient - Allow --meta on upload | 00:39 |
| clayg | timburke: has an answer for everything | 00:42 |
| *** csmart has joined #openstack-swift | 00:50 | |
| timburke | oh man, this is great -- now that i've seen it, i can't *not* hit bug 1736840 | 00:50 |
| openstack | bug 1736840 in OpenStack Object Storage (swift) "Suffix-range requests for 0-byte objects produce invalid responses" [Undecided,In progress] https://launchpad.net/bugs/1736840 - Assigned to Samuel Merritt (torgomatic) | 00:50 |
| timburke | got another, different traceback testing symlinks -- but it's not coming up out of the middleware, so we're better off addressing it separately | 00:51 |
| torgomatic | timburke: there's a partial fix up already; you could see if that helps | 00:51 |
| torgomatic | EC's still busted, but I'm working on it | 00:52 |
| timburke | yeah, it's the ec side that caused the trouble. with a replicated policy for the symlink, it all workd like a champ! | 00:52 |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: Add object count and bytes to ShardRange repr https://review.openstack.org/526561 | 00:54 |
| *** tovin07_ has joined #openstack-swift | 00:55 | |
| clayg | acoles: you're killing it on the deep containers! | 00:59 |
| *** gyee has quit IRC | 01:01 | |
| acoles | clayg: TBH I am mosty salvaging snippets from otherwise abandoned patches | 01:01 |
| clayg | so you're killing... patches... or at least putting them out of their misery? still seems like a noble and worthwhile endevour | 01:02 |
| clayg | kudos! | 01:02 |
| timburke | all right, i think i'm getting it... it boils down to the is_success check in _recursive_get_head -- if you provide a `if-none-match: <empty-md5>` header, the object server responds 304 which makes its way all the way back to the client | 01:03 |
| clayg | timburke: so I'm landing somewhere around here -> https://gist.github.com/clayg/6969949a611add5c004a6d4efe3c6f1f | 01:03 |
| clayg | ... and we can drop all the conditional handling in the proxy & obj server (or so say the unit & func tests) | 01:03 |
| acoles | feels a bit like a reunion party in this channel today :) | 01:04 |
| clayg | but it doesn't really seem like a data model problem... and i'm not even sure it makes a difference during rolling upgrade | 01:04 |
| clayg | my gut is leaning towards follow on patch - but of course I'd be happy to see it squashed | 01:04 |
| clayg | i still need to look into the probe test (which is for container sync, and not failure modes related to split brain while POSTing symlinks sadly) | 01:05 |
| clayg | and I have the worry about the cross account container sync symlinking | 01:05 |
| clayg | sigh - thank you tim - it is *absolutely* a rolling upgrade cruft challange if we don't fix it out the gate | 01:08 |
| clayg | wait... is it | 01:08 |
| clayg | old proxies would be expecting the 200's - so if your storage layer upgrades you're boned | 01:10 |
| timburke | i think i'm liking clayg's gist -- m_kazuhiro and kota_, are you ok with object layer reporting 3xx/4xx statuses, or would you have monitoring that will think something's gone wrong? | 01:27 |
| clayg | how could you possibly infer something bad from a metric that's totally dependent on client behavior? | 01:28 |
| clayg | also... should symlink=get requests not be conditional? why can't I stick a if-modified-since on a symlink=get request? | 01:29 |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Do not leak symlinks to proxy & obj layer https://review.openstack.org/526564 | 01:31 |
| timburke | if i see a spike in *anything* i'd want to investigate ;-) whole bunch of 499s? what's up with that? suddenly i'm getting 10x my normal # of requests? better get some more servers up and figure out if this is supposed to be the new normal | 01:31 |
| timburke | anyway, we could later add some x-backend-* header that says "hey, if you find any of this list of metadata, turn off conditional responses" to clean up the status codes *without* the obj server needing to know about symlinks | 01:32 |
| timburke | symlink=get is *always* conditional | 01:33 |
| clayg | oh oh - yeah I'm not sure why I have such trouble with that boolean there | 01:34 |
| clayg | x-backend-conditional-override-metadata: x-object-sysmeta-symlink-target[, ...] | 01:36 |
| *** m_kazuhiro has joined #openstack-swift | 01:38 | |
| m_kazuhiro | good morning | 01:43 |
| openstackgerrit | Samuel Merritt proposed openstack/swift master: Fix suffix-byte-range responses for zero-byte EC objects. https://review.openstack.org/526565 | 01:48 |
| *** two_tired has joined #openstack-swift | 02:00 | |
| m_kazuhiro | clayg: timburke : You discussed symlink's conditional response on irc. What is your main concerning point? (I'm reading irc dicssion logs now.) | 02:04 |
| kota_ | I'm here at my office | 02:04 |
| clayg | m_kazuhiro: patch 526564 | 02:05 |
| patchbot | https://review.openstack.org/#/c/526564/ - swift - Do not leak symlinks to proxy & obj layer | 02:05 |
| m_kazuhiro | clayg: Thank you. I'm reading the patch now... | 02:06 |
| kota_ | clayg: my quick look, it looks pretty nice. | 02:06 |
| kota_ | let me run unit/func with the patch | 02:07 |
| clayg | hopefully zuul will get back to us on that soon too | 02:07 |
| kota_ | yeah | 02:08 |
| timburke | only functional delta should be in the response codes getting logged, but the *proxy* statuses still reflect what's going out to the client | 02:08 |
| kota_ | btw, does swiftstack stops the third party jobs? I didn't see it appears to the ci results. | 02:08 |
| kota_ | recently | 02:08 |
| kota_ | at first, unittests runs as OK, Ran 5459 tests in 61.143s. *GREAT* | 02:12 |
| clayg | charz: notmyname: can we add a note to check on community CI results? | 02:12 |
| charz | @clayg We were doing VM migration last two weeks. | 02:14 |
| clayg | understood - carry on good sir - thank you | 02:14 |
| charz | kota_: I'll try to bring the CI server ASAP. | 02:15 |
| kota_ | charz: thanks! the information from your CI is really helpful to find the bugs, on the probe in particular | 02:16 |
| clayg | m_kazuhiro: what was the reasoning behind sticking the client x-symlink-target and x-symlink-target-account headers into a single piece of sysmeta? | 02:19 |
| m_kazuhiro | clayg: At implementation time, I thought single piece is simple... | 02:21 |
| m_kazuhiro | clayg: But as you wrote on gerrit, it makes container-sync implementation complexed. | 02:22 |
| *** mat128 has joined #openstack-swift | 02:23 | |
| clayg | yeah... but there's no getting around container-sync knowing about symlinks | 02:24 |
| clayg | I'm just wondering how sad we'll be down the road if we don't support the concept of an explicitly "free" or "accountless" symlink in the data-model | 02:28 |
| clayg | bytes on disk are the worst | 02:28 |
| kota_ | clayg: so if we store x-symlink-target and x-symlink-target-account as separate sysmeta, it's easy to handle what container-sync to transfer? | 02:29 |
| kota_ | just transfer x-symlink-target-account if it exists, e.g. | 02:29 |
| clayg | that was what made me notice... but i'm not sure I have a concern | 02:30 |
| clayg | it's just - you can *tell* if a user said 'x-symlink-target: c/o' or they said 'x-symlink-target: c/o' & 'x-symlink-target-account: a' but a happened to be in the same account they stored the symlink in at the time... | 02:30 |
| clayg | you don't ever really have "accountless" symlinks - but we kind of support them anyway - or at least container-sync does... but implicitly | 02:31 |
| kota_ | true | 02:31 |
| kota_ | let me try to write the sample in today's work? | 02:32 |
| clayg | sorry, i'm not sure it's important yet - just noticing and thinking | 02:32 |
| clayg | I feel a *little* more sure the conditional request thing is a good idea | 02:32 |
| kota_ | hehe | 02:32 |
| clayg | i need to break for dinner tho i'm afraid | 02:32 |
| kota_ | tbh, I and m_kazuhiro had conversation on that, and we're also not sure on that. Good point you noticed ;) | 02:33 |
| clayg | i made much good progress! I think timburke and zaitcev are looking too | 02:33 |
| clayg | we will merge it!! | 02:33 |
| clayg | very soon! | 02:33 |
| clayg | yay!!! | 02:33 |
| clayg | m_kazuhiro: YOU ARE HERO! | 02:33 |
| clayg | the end is in sight | 02:33 |
| m_kazuhiro | clayg: Thank you! | 02:34 |
| kota_ | clayg: one thing | 02:34 |
| kota_ | even we divide the header, could we have the symlink handle (translate from the sysmeta to the user meta for the remote cluster) in the conatiner-sync code? | 02:35 |
| kota_ | i had an opinion, i'd love to make it in because if we do not do so, we should modify the pipeline of internal client that can cause operational failures. | 02:36 |
| *** JimCheung has quit IRC | 02:37 | |
| *** JimCheung has joined #openstack-swift | 02:50 | |
| *** JimCheung has quit IRC | 02:54 | |
| *** links has joined #openstack-swift | 03:42 | |
| charz | kota_: clayg the community CI are back, let me know if you still see problems on it. | 03:43 |
| kota_ | charz: nice, thanks! | 03:43 |
| *** m_kazuhiro has quit IRC | 04:37 | |
| *** mat128 has quit IRC | 04:43 | |
| *** rcernin has quit IRC | 05:11 | |
| *** threestrands has quit IRC | 05:24 | |
| *** threestrands has joined #openstack-swift | 05:39 | |
| *** threestrands has quit IRC | 05:39 | |
| *** threestrands has joined #openstack-swift | 05:39 | |
| *** armaan has joined #openstack-swift | 06:01 | |
| *** armaan has quit IRC | 06:02 | |
| *** rcernin has joined #openstack-swift | 06:11 | |
| *** cshastri has joined #openstack-swift | 06:13 | |
| *** two_tired has quit IRC | 06:23 | |
| *** threestrands has quit IRC | 06:28 | |
| *** cbartz has joined #openstack-swift | 06:29 | |
| *** m_kazuhiro has joined #openstack-swift | 06:52 | |
| -openstackstatus- NOTICE: Due to some unforseen Zuul issues the gate is under very high load and extremely unstable at the moment. This is likely to persist until PST morning | 07:04 | |
| *** ChanServ changes topic to "Due to some unforseen Zuul issues the gate is under very high load and extremely unstable at the moment. This is likely to persist until PST morning" | 07:04 | |
| *** amito has quit IRC | 07:06 | |
| *** seongsoocho has quit IRC | 07:07 | |
| *** serverascode has quit IRC | 07:07 | |
| *** seongsoocho has joined #openstack-swift | 07:07 | |
| *** onovy has quit IRC | 07:08 | |
| *** kencjohnston has quit IRC | 07:08 | |
| *** serverascode has joined #openstack-swift | 07:08 | |
| *** tdasilva has quit IRC | 07:08 | |
| *** cshastri has quit IRC | 07:11 | |
| *** onovy has joined #openstack-swift | 07:11 | |
| *** tdasilva has joined #openstack-swift | 07:13 | |
| *** kencjohnston has joined #openstack-swift | 07:13 | |
| *** csmart has quit IRC | 07:14 | |
| *** armaan has joined #openstack-swift | 07:17 | |
| *** armaan has quit IRC | 07:21 | |
| *** m_kazuhiro has quit IRC | 07:41 | |
| *** hseipp has joined #openstack-swift | 07:53 | |
| *** ukaynar has quit IRC | 07:57 | |
| *** ukaynar has joined #openstack-swift | 07:58 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: More cleanup for Clay's great work https://review.openstack.org/526614 | 07:58 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: Testing Implementation to devide symlink headers https://review.openstack.org/526615 | 07:58 |
| *** ukaynar has quit IRC | 08:03 | |
| *** tesseract has joined #openstack-swift | 08:20 | |
| *** armaan has joined #openstack-swift | 08:45 | |
| *** armaan has quit IRC | 08:53 | |
| *** armaan has joined #openstack-swift | 08:55 | |
| *** armaan has quit IRC | 08:57 | |
| *** itlinux has quit IRC | 09:01 | |
| *** hseipp has quit IRC | 09:05 | |
| *** amito has joined #openstack-swift | 09:06 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: Testing Implementation to devide symlink headers https://review.openstack.org/526615 | 09:07 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: Testing Implementation to devide symlink headers https://review.openstack.org/526615 | 09:13 |
| kota_ | clayg, timburke: for the conditional response of symlink, i like the clay's patch. then, probably https://review.openstack.org/#/c/526614/ is better to squash in. | 09:15 |
| patchbot | patch 526614 - swift - More cleanup for Clay's great work | 09:15 |
| kota_ | and https://review.openstack.org/#/c/526615/ is my trying to separate the symlink target sysmeta to cont+object and account. | 09:16 |
| patchbot | patch 526615 - swift - Testing Implementation to devide symlink headers | 09:16 |
| kota_ | it looks to work well but i need to circle back again to check though the whole patch set. | 09:17 |
| kota_ | whole of the patch. | 09:17 |
| kota_ | it makes container-sync change smaller so that you may like it and i'm fine to change the symlink to do so. | 09:17 |
| kota_ | i'll up tomorrow too to keep track the discussion so that it'll be happy you would check them while i asleep. | 09:18 |
| kota_ | tomorrow, it's weekend though, i have to do small things (release) for storlets anway, bercause zuul is too slow today :,( | 09:19 |
| kota_ | if you find anything, please let me (or m_kazuhiro) know. | 09:19 |
| kota_ | clayg, timburke, zaitcev: ^^^^ | 09:20 |
| kota_ | thx | 09:20 |
| *** csmart has joined #openstack-swift | 09:21 | |
| *** cshastri has joined #openstack-swift | 09:26 | |
| *** hseipp has joined #openstack-swift | 09:47 | |
| *** cbartz has quit IRC | 09:48 | |
| *** ukaynar has joined #openstack-swift | 09:58 | |
| *** ukaynar has quit IRC | 10:03 | |
| *** cbartz has joined #openstack-swift | 10:04 | |
| *** tovin07_ has quit IRC | 10:23 | |
| *** hoonetorg has quit IRC | 10:44 | |
| *** hoonetorg has joined #openstack-swift | 10:45 | |
| *** openstackgerrit has quit IRC | 11:17 | |
| *** links has quit IRC | 11:28 | |
| *** kei_yama has quit IRC | 11:37 | |
| *** links has joined #openstack-swift | 11:41 | |
| *** tesseract has quit IRC | 11:50 | |
| *** tesseract has joined #openstack-swift | 11:51 | |
| *** silor has joined #openstack-swift | 11:57 | |
| *** ukaynar has joined #openstack-swift | 11:59 | |
| *** ukaynar has quit IRC | 12:04 | |
| *** cshastri has quit IRC | 12:34 | |
| *** mat128 has joined #openstack-swift | 13:16 | |
| *** mat128 has quit IRC | 13:24 | |
| *** openstackgerrit has joined #openstack-swift | 13:30 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: Testing Implementation to devide symlink headers https://review.openstack.org/526615 | 13:30 |
| *** mat128 has joined #openstack-swift | 13:30 | |
| *** mat128 has quit IRC | 13:40 | |
| *** tesseract has quit IRC | 13:47 | |
| *** tesseract has joined #openstack-swift | 13:50 | |
| *** ukaynar has joined #openstack-swift | 14:00 | |
| *** ChanServ changes topic to "https://bugs.not.mn/project/Swift | Logs: http://eavesdrop.openstack.org/irclogs/%23openstack-swift/ | Meetings: https://wiki.openstack.org/wiki/Meetings/Swift | Review Dashboard: http://not.mn/reviews.html" | 14:01 | |
| -openstackstatus- NOTICE: The issues have been fixed, Zuul is operating fine again but has a large backlog. You can recheck jobs that failed. | 14:01 | |
| *** ukaynar has quit IRC | 14:04 | |
| *** mat128 has joined #openstack-swift | 14:05 | |
| *** mat128 has quit IRC | 14:05 | |
| *** mat128 has joined #openstack-swift | 14:35 | |
| *** links has quit IRC | 14:37 | |
| *** vinsh has quit IRC | 14:46 | |
| *** rcernin has quit IRC | 14:55 | |
| zaitcev | What's "devide" in this case? "Decide"? "Divide"? | 15:10 |
| *** hseipp has quit IRC | 15:40 | |
| openstackgerrit | Merged openstack/swift feature/deep: Use the same hash and created_at value in shard db https://review.openstack.org/516488 | 15:43 |
| *** hseipp has joined #openstack-swift | 15:45 | |
| *** cbartz has quit IRC | 15:58 | |
| *** ukaynar has joined #openstack-swift | 15:59 | |
| *** itlinux has joined #openstack-swift | 16:14 | |
| *** mat128 has quit IRC | 16:27 | |
| *** zaitcev has quit IRC | 16:33 | |
| *** itlinux has quit IRC | 16:42 | |
| *** itlinux has joined #openstack-swift | 16:43 | |
| *** zaitcev has joined #openstack-swift | 16:46 | |
| *** ChanServ sets mode: +v zaitcev | 16:46 | |
| timburke | on p 526615 -- what do we want the behavior to be for cross-account copies? given a valid symlink in one account, i guess copying it (with a ?symlink=get) should produce different results depending upon which headers were sent when the symlink was PUT? | 16:50 |
| patchbot | https://review.openstack.org/#/c/526615/ - swift - Testing Implementation to devide symlink headers | 16:50 |
| timburke | it feels a little weird to me though... like it's an extra layer of complication for clients -- they have to do a ?symlink=get GET *first* before they can know what effect the COPY will have, at which point they may as well do the PUT themselves | 16:50 |
| *** d0ugal has quit IRC | 16:52 | |
| zaitcev | Yea, I'm looking at it too. | 16:52 |
| zaitcev | In fact, when I PUT, my supplied account header goes straight into sysmeta (in both implementations). Then, as I do GET, it gets into the traversal request now. Doesn't it allow me to peek into ANY account? | 16:53 |
| *** d0ugal has joined #openstack-swift | 16:54 | |
| *** d0ugal has quit IRC | 16:54 | |
| *** d0ugal has joined #openstack-swift | 16:54 | |
| timburke | kind of. but the request is (or at least, should be, *needs to be*) re-authed as we follow redirects. you can create a symlink that points somewhere you can't access (or doesn't exist), but trying to follow it should 403 (or 404) | 16:55 |
| timburke | at least, that's what my mental model says :-) | 16:56 |
| zaitcev | I need to look closer. | 16:56 |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: Add object count and bytes to ShardRange repr https://review.openstack.org/526561 | 16:56 |
| *** itlinux has quit IRC | 17:00 | |
| *** itlinux has joined #openstack-swift | 17:05 | |
| *** gyee has joined #openstack-swift | 17:14 | |
| *** JimCheung has joined #openstack-swift | 17:19 | |
| *** hseipp has quit IRC | 17:27 | |
| *** tesseract has quit IRC | 17:29 | |
| *** ukaynar has quit IRC | 17:32 | |
| clayg | heyoh! | 17:33 |
| *** ukaynar has joined #openstack-swift | 17:41 | |
| *** openstackgerrit has quit IRC | 18:03 | |
| *** openstackgerrit has joined #openstack-swift | 18:21 | |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: Add probe test to delete then revive a sharded container https://review.openstack.org/526552 | 18:21 |
| clayg | timburke: it seems like patch 476992 should have a closes-bug associated with it? | 18:59 |
| patchbot | https://review.openstack.org/#/c/476992/ - swift - Make If-None-Match:* work properly with 0-byte PUTs | 18:59 |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Do not leak symlinks to proxy & obj layer https://review.openstack.org/526564 | 19:10 |
| timburke | clayg: yeah, probably should write up a bug for it... | 19:11 |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Testing Implementation to divide symlink headers https://review.openstack.org/526615 | 19:12 |
| *** __david has joined #openstack-swift | 19:26 | |
| clayg | zaitcev: no worries then - i probably had extra context from discussing the patch in channel with kota... | 19:38 |
| clayg | zaitcev: it's definitely divide - it takes the x-symlink-target and x-symlink-target-account and divide's them into *separate* metadata instead of cramming them into the single absolute path. | 19:39 |
| zaitcev | Oh. I see... I thought the division was about subdividing sysmeta and meta, but I wasn't sure. | 19:42 |
| openstackgerrit | Merged openstack/python-swiftclient master: Allow --meta on upload https://review.openstack.org/481202 | 19:51 |
| *** ChanServ has quit IRC | 20:17 | |
| *** ChanServ has joined #openstack-swift | 20:24 | |
| *** barjavel.freenode.net sets mode: +o ChanServ | 20:24 | |
| clayg | I'm pondering versioned writes + symlinks | 20:25 |
| clayg | timburke: I think you've advocated most strongly for the symlink_path in the container listings... but it's causing me some worry on patch 232162 and patch 526615' | 20:26 |
| patchbot | https://review.openstack.org/#/c/232162/ - swift - Symlink implementation. | 20:26 |
| patchbot | https://review.openstack.org/#/c/526615/ - swift - Testing Implementation to divide symlink headers | 20:26 |
| timburke | clayg: my main reason for it is just that, if we don't do it now, *we'll never have it, ever* -- see the resistance (or at any rate, apathy) to https://review.openstack.org/#/c/337960/ | 20:32 |
| patchbot | patch 337960 - swift - Send correct SLO ETag for container updates | 20:32 |
| *** itlinux has quit IRC | 20:55 | |
| *** itlinux has joined #openstack-swift | 20:59 | |
| *** JimCheung has quit IRC | 21:01 | |
| *** JimCheung has joined #openstack-swift | 21:01 | |
| *** ukaynar has quit IRC | 21:31 | |
| *** ukaynar has joined #openstack-swift | 21:32 | |
| *** ukaynar has quit IRC | 21:35 | |
| *** ukaynar has joined #openstack-swift | 21:35 | |
| *** silor has quit IRC | 21:36 | |
| clayg | timburke: sure... but in this case it's causing some specific complications... (leakage if symlinks is removed & ambiguity of absolute and relative symlinks) | 21:43 |
| *** rcernin has joined #openstack-swift | 21:44 | |
| clayg | in case anyone else is totally psyched about this - python-swiftclient `swift post <cont> <symlink>` automatically follows the 307 and you end up applying the metadata to BOTH which is basically the ideal behavior. It's amazing when HTTP clients and services do the right thing. | 21:58 |
| clayg | it's almost as if this whole "RESTful interface" isn't total bullshit ๐ | 21:59 |
| *** itlinux has quit IRC | 22:01 | |
| *** rcernin has quit IRC | 22:03 | |
| timburke | whooo! our requests dependency actually worked in our favor! | 22:03 |
| *** rcernin has joined #openstack-swift | 22:03 | |
| *** ukaynar has quit IRC | 22:07 | |
| *** ukaynar has joined #openstack-swift | 22:08 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift master: Fix small error in a doc string https://review.openstack.org/526791 | 22:09 |
| clayg | timburke: in VW, in history mode - what does one *do* with the zero byte marker? | 22:09 |
| clayg | https://gist.github.com/clayg/0103ef60243c0a174002c4e1b8c9e566 | 22:11 |
| timburke | it serves as a record of the fact that there was a client request to delete that key | 22:12 |
| clayg | and what is `swift_versions_deleted=1` all about? it's not mentioned on https://docs.openstack.org/swift/latest/overview_object_versioning.html | 22:13 |
| timburke | it's something only middleware can set, because of the `;swift`-ness of it. so we *know* it came from versioned_writes | 22:15 |
| timburke | on the multiple-delete behavior, i was largely cribbing behavior from http://docs.aws.amazon.com/AmazonS3/latest/dev/RemDelMarker.html | 22:15 |
| *** saint_ has quit IRC | 22:16 | |
| *** itlinux has joined #openstack-swift | 22:16 | |
| *** ianychoi_ has joined #openstack-swift | 22:17 | |
| *** itlinux has quit IRC | 22:18 | |
| *** ianychoi has quit IRC | 22:19 | |
| kota_ | zaitcev: divide, i meant. Sorry | 22:33 |
| * kota_ is looking at log | 22:34 | |
| clayg | damn you flakey tests; lp bug #1724356 | 22:35 |
| openstack | Launchpad bug 1724356 in OpenStack Object Storage (swift) "test_part_swapping_problem is flakey" [Medium,Confirmed] https://launchpad.net/bugs/1724356 | 22:35 |
| clayg | bummer - I can't figure out how to create a symlink with swift cli - for zero-byte object I normally do `echo | swift upload test - --object-name zero` | 22:44 |
| clayg | but that results in chunked upload that just happens to not send any bytes - and the symlink middleware explcitly wants `content_length == 0` for PUT | 22:45 |
| clayg | I tried to add the header - but it seems to throw off swiftclients guts... `sendall() argument 1 must be convertible to a buffer, not ReadableToIterable` too bad | 22:45 |
| timburke | wait, what's that last part? i wanna see a traceback! | 23:09 |
| timburke | for the first, have you tried just touching a file? or uploading /dev/null? | 23:09 |
| clayg | timburke: https://gist.github.com/clayg/2a713f1b7a5b3cc44973411579db5b21 | 23:16 |
| zaitcev | kota_: Clay has set me straight. Truth to be told, I tried to discuss "ใใ้ใฎในในใก" a few days ago... the symbols look too similar! So yeah | 23:44 |
| kota_ | That's nice. Thanks clayg | 23:47 |
| kota_ | zaitcev: make sense, i googled ใใๅ ใฎในในใก | 23:49 |
| *** ukaynar has quit IRC | 23:49 | |
| kota_ | symbol is similar but the mean is completely different | 23:49 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!