| timburke | ok, i'm gonna go run through p 532865 and actually set up barbican & keystone, with an eye toward +2 so we can fix the broken. wish me luck! | 00:29 |
|---|---|---|
| patchbot | https://review.openstack.org/#/c/532865/ - swift - fix barbican integration | 00:29 |
| *** ukaynar has quit IRC | 00:42 | |
| *** ukaynar has joined #openstack-swift | 00:43 | |
| mattoliverau | timburke: good luck :) | 00:43 |
| *** ukaynar has quit IRC | 00:47 | |
| *** armaan has joined #openstack-swift | 01:00 | |
| *** ukaynar has joined #openstack-swift | 01:00 | |
| *** tovin07_ has joined #openstack-swift | 01:01 | |
| *** armaan has quit IRC | 01:06 | |
| *** bigdogstl has quit IRC | 01:14 | |
| *** tovin07 has quit IRC | 01:22 | |
| *** tovin07 has joined #openstack-swift | 01:22 | |
| clayg | timburke: as I look at patch 534959 I'm reminded just how weird the expirer is about handling overwrites... if you find one node to accept the expiration of t1.data with an x-delete-at.ts the object is gone (eventually) even if a t2.data exists out there with a different x-delete-at | 01:26 |
| patchbot | https://review.openstack.org/#/c/534959/ - swift - object-server can 409 in response to x-if-delete-at | 01:26 |
| clayg | I remember one time thinking about using the x-timestamp (plus an offset) of the .data that was to-be-delete as the timestamp of the tombstone (i.e. delete t1.data with t1_1.ts so any later writes trump the t1_1.ts implicitly - but there was a problem with writing tombstones in the past getting reclaimed too quickly/immediately) | 01:28 |
| *** SkyRocknRoll has quit IRC | 01:35 | |
| openstackgerrit | Merged openstack/python-swiftclient master: Revert "Add Constraints support" https://review.openstack.org/534881 | 01:37 |
| openstackgerrit | Merged openstack/python-swiftclient master: Allow for object uploads > 5GB from stdin. https://review.openstack.org/478611 | 01:37 |
| *** bigdogstl has joined #openstack-swift | 01:45 | |
| *** vint_bra has joined #openstack-swift | 01:46 | |
| *** SkyRocknRoll has joined #openstack-swift | 01:47 | |
| *** geaaru__ has joined #openstack-swift | 01:48 | |
| timburke | clayg: yeah -- it's tough. even using the x-delete-at timestamp is risky, depending on how far behind your expirers are running | 01:49 |
| timburke | but at least we cover pop_queue failures and overwrites that come *after* the x-delete-at (but presumably still within a reclaim_age) | 01:50 |
| *** bigdogstl has quit IRC | 01:50 | |
| *** geaaru_ has quit IRC | 01:50 | |
| *** m_kazuhiro has joined #openstack-swift | 01:50 | |
| m_kazuhiro | good morning | 01:51 |
| timburke | clayg: note that there's a parent patch: https://review.openstack.org/#/c/534399/ | 01:51 |
| patchbot | patch 534399 - swift - internal_client: Don't retry when we expect the sa... | 01:51 |
| clayg | timburke: yeah i just started on that one | 01:52 |
| timburke | thanks! | 01:52 |
| *** bigdogstl has joined #openstack-swift | 01:57 | |
| *** ukaynar has quit IRC | 01:59 | |
| *** ukaynar has joined #openstack-swift | 01:59 | |
| kota_ | good morning | 02:00 |
| *** kukacz_ has quit IRC | 02:00 | |
| *** kukacz_ has joined #openstack-swift | 02:01 | |
| *** vint_bra has quit IRC | 02:01 | |
| *** ukaynar has quit IRC | 02:03 | |
| *** ukaynar has joined #openstack-swift | 02:04 | |
| mattoliverau | m_kazuhiro, kota_: morning | 02:06 |
| *** ukaynar has quit IRC | 02:09 | |
| *** bigdogstl has quit IRC | 02:09 | |
| clayg | timburke: kota_: did an earlier version of patch 534399 already try adding 409 to "server_handled_successfully" | 02:11 |
| patchbot | https://review.openstack.org/#/c/534399/ - swift - internal_client: Don't retry when we expect the sa... | 02:11 |
| m_kazuhiro | mattoliverau: morning! Will you go to the Dublin PTG? If you go, I want to know which day you are in Dublin. | 02:12 |
| mattoliverau | m_kazuhiro: Yes, I'll be there. I'm arriving on Sunday afternoon and then on Friday night flying to edinburgh (scotland) to spend the weekend with my Brother before I fly home on the Monday :) | 02:18 |
| *** vint_bra has joined #openstack-swift | 02:20 | |
| *** vint_bra has quit IRC | 02:23 | |
| m_kazuhiro | mattoliverau: Great! So, you can have discussion time from monday to friday AM, right? | 02:25 |
| mattoliverau | m_kazuhiro: yup :) | 02:25 |
| m_kazuhiro | mattoliverau: ok. Because I want to merge expirer's general task queue patch at the PTG, I want to get discussion time with you at Dublin. | 02:28 |
| mattoliverau | m_kazuhiro: sure, and then we can start working on the refactors ;) | 02:29 |
| kota_ | clayg: sorry, what does that mean? Do you want to add 409 fix for reconciler in the patch? | 02:31 |
| m_kazuhiro | mattoliverau: Yes, I need to refactor the patch :) | 02:32 |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Consider 409 not a server error; always https://review.openstack.org/535575 | 02:33 |
| kota_ | clayg: oh, that's what you considered. | 02:36 |
| kota_ | clayg: and i just kicked off the gate job. if you mind it, it's ok you could halt the job via sort of -2. | 02:37 |
| kota_ | i think it's the way to stop the merge jobs. | 02:38 |
| clayg | nah, it's fine - timburke will probably explain why patch 535575 is a bad idea and entirely insufficient while his more complicated special casing is absolute necessary | 02:38 |
| patchbot | https://review.openstack.org/#/c/535575/ - swift - Consider 409 not a server error; always | 02:38 |
| clayg | then i can just abandon it and the right code will already be on master | 02:39 |
| *** bigdogstl has joined #openstack-swift | 02:40 | |
| kota_ | In my understanding, the 409 would be changed if the request timestamp is changed. e.g. if a newer object found when delete request incomming, it will be 409 but re-trying w/o same x-stimestamp will cause 204 because the timestamp will be increased. | 02:41 |
| kota_ | and the logci in the p 534399 is in the make_request method so that, in general, 409 and the x-timestamp supplied will be consistent result (and safe to give up). | 02:43 |
| patchbot | https://review.openstack.org/#/c/534399/ - swift - internal_client: Don't retry when we expect the sa... | 02:43 |
| kota_ | if x-timestamp is not set, the result of the request re-try would be changed. | 02:43 |
| *** bigdogstl has quit IRC | 02:45 | |
| *** threestrands_ has joined #openstack-swift | 02:49 | |
| *** threestrands_ has quit IRC | 02:49 | |
| *** threestrands_ has joined #openstack-swift | 02:49 | |
| *** threestrands has quit IRC | 02:51 | |
| *** ___david has quit IRC | 02:52 | |
| *** two_tired has joined #openstack-swift | 02:54 | |
| *** bigdogstl has joined #openstack-swift | 03:03 | |
| *** bigdogstl has quit IRC | 03:13 | |
| openstackgerrit | Merged openstack/swift master: Return HTTPServerError instead of HTTPNotFound https://review.openstack.org/533492 | 03:26 |
| *** psachin has joined #openstack-swift | 03:31 | |
| *** bigdogstl has joined #openstack-swift | 03:37 | |
| timburke | clayg: no -- i was being conservative with the change. i could definitely see a 409 that could succeed with a retry, though | 03:42 |
| timburke | two simultaneous writers with slightly different clocks, for example | 03:42 |
| clayg | A 404 *could* succeed with a retry. | 03:43 |
| timburke | of course, if the caller provided a timestamp, there's no way that it would | 03:43 |
| timburke | hell, we retry *400s* | 03:44 |
| *** jappleii__ has joined #openstack-swift | 03:44 | |
| timburke | like, *maybe* that'll help in an upgrade? who the hell knows... but if it *did* manage to help, i feel like it's just papering over some bad | 03:44 |
| *** jappleii__ has quit IRC | 03:45 | |
| *** jappleii__ has joined #openstack-swift | 03:45 | |
| timburke | it's almost certainly better to only retry socket errors and 5xx | 03:45 |
| timburke | (likely, we could even carve out an exception to *that* for 501) | 03:46 |
| *** jappleii__ has quit IRC | 03:46 | |
| *** threestrands_ has quit IRC | 03:46 | |
| *** jappleii__ has joined #openstack-swift | 03:47 | |
| *** jappleii__ has quit IRC | 03:48 | |
| *** jappleii__ has joined #openstack-swift | 03:48 | |
| *** rcernin has quit IRC | 03:49 | |
| *** rcernin has joined #openstack-swift | 03:49 | |
| *** bigdogstl has quit IRC | 03:51 | |
| mattoliverau | https://gist.github.com/matthewoliver/ce5172dbf1bcf8143184686517ad67ca <-- a new version of a part diff that takes 2 builders or rings. And if you tell it'll print out what parts are different in table form.. because apparently I'm easy distracted exp while this barbican env builds. NOTE: a -v on a big rings will print alot (the whole replica2part2dev) so not sure how usefult it really is, but was fun to write :) | 03:55 |
| *** kei_yama has quit IRC | 03:56 | |
| openstackgerrit | ShangXiao proposed openstack/swift master: Put instructions for the response codes https://review.openstack.org/535592 | 03:58 |
| *** kei_yama has joined #openstack-swift | 04:01 | |
| *** SkyRocknRoll_ has joined #openstack-swift | 04:11 | |
| openstackgerrit | Merged openstack/swift master: internal_client: Don't retry when we expect the same reponse https://review.openstack.org/534399 | 04:12 |
| *** bigdogstl has joined #openstack-swift | 04:17 | |
| *** bigdogstl has quit IRC | 04:22 | |
| *** bigdogstl has joined #openstack-swift | 04:24 | |
| *** geaaru_ has joined #openstack-swift | 04:25 | |
| *** geaaru__ has quit IRC | 04:28 | |
| *** bigdogstl has quit IRC | 04:29 | |
| *** bigdogstl has joined #openstack-swift | 04:35 | |
| *** links has joined #openstack-swift | 04:37 | |
| *** m_kazuhiro has quit IRC | 04:52 | |
| *** two_tired has quit IRC | 04:55 | |
| *** bigdogstl has quit IRC | 04:57 | |
| *** SkyRocknRoll_ has quit IRC | 05:15 | |
| *** gkadam has joined #openstack-swift | 05:28 | |
| *** SkyRocknRoll_ has joined #openstack-swift | 05:43 | |
| *** jappleii__ has quit IRC | 05:52 | |
| *** bigdogstl has joined #openstack-swift | 05:56 | |
| *** cshastri_ has joined #openstack-swift | 05:57 | |
| *** f0o has quit IRC | 06:07 | |
| *** bigdogstl has quit IRC | 06:07 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift feature/s3api: Merge remote-tracking branch 'remotes/origin/master' into merge-master https://review.openstack.org/535623 | 06:13 |
| *** f0o has joined #openstack-swift | 06:15 | |
| *** itlinux has joined #openstack-swift | 06:16 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift feature/s3api: WIP: merge func tests https://review.openstack.org/532545 | 06:20 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift feature/s3api: As a first step, let's skip the test if not s3api in the middleware https://review.openstack.org/534703 | 06:20 |
| *** armaan has joined #openstack-swift | 06:29 | |
| *** cshastri_ has quit IRC | 06:34 | |
| *** pcaruana has joined #openstack-swift | 06:44 | |
| *** geaaru__ has joined #openstack-swift | 06:57 | |
| *** geaaru_ has quit IRC | 06:59 | |
| *** bigdogstl has joined #openstack-swift | 07:02 | |
| *** gkadam has quit IRC | 07:03 | |
| *** geaaru__ has quit IRC | 07:06 | |
| *** bigdogstl has quit IRC | 07:14 | |
| *** bigdogstl has joined #openstack-swift | 07:25 | |
| *** itlinux has quit IRC | 07:29 | |
| *** bigdogstl has quit IRC | 07:30 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift feature/s3api: Let's skip the test if not s3api in the middleware https://review.openstack.org/534703 | 07:36 |
| *** bigdogstl has joined #openstack-swift | 07:37 | |
| *** rcernin has quit IRC | 07:43 | |
| *** bigdogstl has quit IRC | 07:44 | |
| openstackgerrit | yanwei proposed openstack/swift master: Add license comment of packages to requirements.txt and test-requirements.txt https://review.openstack.org/535655 | 07:53 |
| openstackgerrit | yanwei proposed openstack/swift master: Add license comment of packages to requirements.txt and test-requirements.txt https://review.openstack.org/535655 | 08:00 |
| openstackgerrit | yanwei proposed openstack/swift master: Add license comments of packages to *requirements.txt https://review.openstack.org/535655 | 08:03 |
| *** bigdogstl has joined #openstack-swift | 08:10 | |
| *** tesseract has joined #openstack-swift | 08:16 | |
| *** geaaru has joined #openstack-swift | 08:19 | |
| *** bigdogstl has quit IRC | 08:20 | |
| *** tovin07__ has joined #openstack-swift | 08:25 | |
| *** flwang has quit IRC | 08:28 | |
| *** StevenK has quit IRC | 08:28 | |
| *** tovin07_ has quit IRC | 08:29 | |
| *** tovin07 has quit IRC | 08:29 | |
| *** StevenK has joined #openstack-swift | 08:29 | |
| *** gkadam has joined #openstack-swift | 08:29 | |
| *** tovin07 has joined #openstack-swift | 08:29 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift feature/s3api: WIP: multipart doesn't work yet other tests run successfully in my local https://review.openstack.org/535667 | 08:35 |
| *** PagliaccisCloud has quit IRC | 08:38 | |
| *** flwang has joined #openstack-swift | 08:44 | |
| *** PagliaccisCloud has joined #openstack-swift | 08:45 | |
| openstackgerrit | yanwei proposed openstack/swift master: Add Docstrings to validate_replicas_by_tier https://review.openstack.org/535675 | 08:58 |
| openstackgerrit | yanwei proposed openstack/swift master: Add Docstrings to validate_replicas_by_tier https://review.openstack.org/535675 | 08:59 |
| *** cshastri_ has joined #openstack-swift | 09:00 | |
| *** bigdogstl has joined #openstack-swift | 09:13 | |
| acoles | good morning | 09:16 |
| mattoliverau | acoles: morning | 09:21 |
| *** cshastri_ has quit IRC | 09:23 | |
| *** bigdogstl has quit IRC | 09:24 | |
| acoles | mattoliverau: team meeting time for you? | 09:25 |
| mattoliverau | acoles: yup, it's Friday evening.. so of course :P | 09:25 |
| acoles | mattoliverau: yeah...it's Friday again...already! | 09:25 |
| mattoliverau | thanks for working on the shardranges in the body.. looks great. Am running probe tests... just in case ;) | 09:26 |
| acoles | mattoliverau: probe tests are the acid test on this branch. func tests aren't going to help us. | 09:44 |
| acoles | there's probably some (a lot!) unit test coverage needed on sharder.py, I'm trying to leave TODO's where I am aware they are needed | 09:45 |
| mattoliverau | yeah thanks, sorry I probably should have been better at writing tests | 09:46 |
| acoles | mattoliverau: no, it's all cool, there's a time for beefing up the tests as the design settles, which is why I'm still leaving some TODO's vs writing them now. Depends on context. | 09:49 |
| *** pcaruana has quit IRC | 09:50 | |
| acoles | for example today I have tedious test rework to do on p 535466 :( | 09:50 |
| patchbot | https://review.openstack.org/#/c/535466/ - swift (feature/deep) - WIP Use backend header rather than param to GET sh... | 09:50 |
| *** bigdogstl has joined #openstack-swift | 09:52 | |
| acoles | mattoliverau: have a good weekend and enjoy the conf | 09:52 |
| *** pcaruana has joined #openstack-swift | 09:53 | |
| *** pcaruana has quit IRC | 09:53 | |
| *** pcaruana has joined #openstack-swift | 09:54 | |
| *** SkyRocknRoll has quit IRC | 09:56 | |
| *** bigdogstl has quit IRC | 10:01 | |
| mattoliverau | acoles: thanks. I hope one day you'll come to an LCA :) | 10:02 |
| mattoliverau | meetings over so calling it a night. | 10:02 |
| acoles | g'night | 10:02 |
| mattoliverau | have a great weekend all o/ | 10:02 |
| *** tovin07__ has quit IRC | 10:06 | |
| *** tovin07 has quit IRC | 10:10 | |
| *** SkyRocknRoll has joined #openstack-swift | 10:13 | |
| *** bigdogstl has joined #openstack-swift | 11:03 | |
| *** bigdogstl has quit IRC | 11:14 | |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: Define singletons to represent min and max shard range bounds https://review.openstack.org/533200 | 11:21 |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: Simplify proxy-server getting sharded listings https://review.openstack.org/533201 | 11:21 |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: only make single request for container listings https://review.openstack.org/512756 | 11:21 |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: enable single PUT of multiple shard ranges to container servers https://review.openstack.org/535407 | 11:21 |
| openstackgerrit | Alistair Coles proposed openstack/swift feature/deep: Use backend header rather than param to GET shard record type https://review.openstack.org/535466 | 11:21 |
| *** cbartz has joined #openstack-swift | 11:21 | |
| *** SkyRocknRoll_ has quit IRC | 11:23 | |
| openstackgerrit | Merged openstack/swift master: fix barbican integration https://review.openstack.org/532865 | 11:40 |
| *** bigdogstl has joined #openstack-swift | 12:00 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift master: Updated from global requirements https://review.openstack.org/88736 | 12:19 |
| *** torgomatic has quit IRC | 12:25 | |
| *** jrichli has quit IRC | 12:26 | |
| *** hogepodge has quit IRC | 12:26 | |
| *** armaan_ has joined #openstack-swift | 12:26 | |
| *** jrichli_ has joined #openstack-swift | 12:27 | |
| *** PagliaccisCloud has quit IRC | 12:27 | |
| *** armaan has quit IRC | 12:27 | |
| *** d34dh0r53 has quit IRC | 12:27 | |
| *** torgomatic has joined #openstack-swift | 12:27 | |
| *** ChanServ sets mode: +v torgomatic | 12:27 | |
| *** d34dh0r53 has joined #openstack-swift | 12:28 | |
| *** hogepodge has joined #openstack-swift | 12:28 | |
| *** PagliaccisCloud has joined #openstack-swift | 12:30 | |
| *** bigdogstl has quit IRC | 12:32 | |
| *** psachin has quit IRC | 12:50 | |
| *** bigdogstl has joined #openstack-swift | 13:02 | |
| *** kei_yama has quit IRC | 13:04 | |
| *** szaher has quit IRC | 13:07 | |
| *** bigdogstl has quit IRC | 13:17 | |
| *** links has quit IRC | 13:28 | |
| openstackgerrit | Merged openstack/swift feature/s3api: Merge remote-tracking branch 'remotes/origin/master' into merge-master https://review.openstack.org/535623 | 13:29 |
| *** bigdogstl has joined #openstack-swift | 13:35 | |
| *** bigdogstl has quit IRC | 13:40 | |
| *** bigdogstl has joined #openstack-swift | 13:42 | |
| *** bigdogstl has quit IRC | 13:46 | |
| *** bigdogstl has joined #openstack-swift | 13:47 | |
| *** bigdogstl has quit IRC | 14:02 | |
| *** bigdogstl has joined #openstack-swift | 14:20 | |
| *** bigdogstl has quit IRC | 14:25 | |
| *** jistr is now known as jistr|mtg | 14:30 | |
| *** bigdogstl has joined #openstack-swift | 14:51 | |
| *** bigdogstl has quit IRC | 14:56 | |
| *** SkyRocknRoll_ has joined #openstack-swift | 14:59 | |
| *** jistr|mtg is now known as jistr | 15:04 | |
| *** bigdogstl has joined #openstack-swift | 15:04 | |
| *** bigdogstl has quit IRC | 15:09 | |
| *** bigdogstl has joined #openstack-swift | 15:13 | |
| *** bigdogstl has quit IRC | 15:29 | |
| *** bigdogstl has joined #openstack-swift | 15:39 | |
| *** bigdogstl has quit IRC | 15:44 | |
| *** kukacz_ has quit IRC | 15:58 | |
| *** kukacz has joined #openstack-swift | 15:59 | |
| *** bigdogstl has joined #openstack-swift | 16:03 | |
| *** jlvillal has quit IRC | 16:04 | |
| *** bigdogstl has quit IRC | 16:07 | |
| *** jlvillal has joined #openstack-swift | 16:14 | |
| *** itlinux has joined #openstack-swift | 16:16 | |
| *** bigdogstl has joined #openstack-swift | 16:18 | |
| *** bigdogstl has quit IRC | 16:23 | |
| *** itlinux has quit IRC | 16:34 | |
| *** itlinux has joined #openstack-swift | 16:40 | |
| *** bigdogstl has joined #openstack-swift | 16:41 | |
| *** bigdogstl has quit IRC | 16:52 | |
| *** gyee has joined #openstack-swift | 16:58 | |
| notmyname | good morning | 17:10 |
| *** SkyRocknRoll_ has quit IRC | 17:18 | |
| *** chsc has joined #openstack-swift | 17:20 | |
| *** chsc has joined #openstack-swift | 17:20 | |
| timburke | good morning | 17:22 |
| *** gkadam has quit IRC | 17:29 | |
| *** cbartz has quit IRC | 17:31 | |
| *** pcaruana has quit IRC | 17:36 | |
| clayg | @timburke well but why isn't server_handled_succssfully just not is_server_error??? | 17:39 |
| clayg | there's a statsd helper/decorator called timing_stats - the tests for it expected a 401 response to be an "error" metric | 17:40 |
| clayg | which I could see an increase in 401's being *maybe* a bad acting client - but it could also mean something is wrong with auth | 17:40 |
| *** bigdogstl has joined #openstack-swift | 17:44 | |
| *** SkyRocknRoll has quit IRC | 17:44 | |
| *** mhks has joined #openstack-swift | 17:47 | |
| *** bigdogstl has quit IRC | 17:51 | |
| timburke | clayg: that's an excellent question :-) | 18:03 |
| *** bigdogstl has joined #openstack-swift | 18:09 | |
| clayg | redbo: any insight on 3586f829 - why only special case 404? why would any 4XX be an "error"? | 18:10 |
| clayg | torgomatic: any insight on on 48d94d96 - why increase special case for 412/416 but not 409/401? | 18:11 |
| redbo | I don't even remember doing that. | 18:12 |
| *** bigdogstl has quit IRC | 18:13 | |
| *** bigdogstl has joined #openstack-swift | 18:17 | |
| clayg | redbo: sure... but that's what the diff is for - to jog your memory | 18:18 |
| clayg | anyway there's a good chance with out some insight from the past timburke and I are going to stop calling 401's "errors" in statsd metrics - so hope that's cool with everyone! | 18:19 |
| timburke | or we have internal_client start limiting retries to just is_server_error() | 18:20 |
| clayg | timburke: sure | 18:21 |
| clayg | timburke: good thinking acctually | 18:21 |
| clayg | timburke: let's do that so redbo and torgomatic don't have to load a bunch of context | 18:21 |
| clayg | i don't really want to mess with statsd anyway | 18:21 |
| *** bigdogstl has quit IRC | 18:22 | |
| *** bigdogstl has joined #openstack-swift | 18:23 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Internal Client should only retry server errors https://review.openstack.org/535575 | 18:26 |
| *** bigdogstl has quit IRC | 18:28 | |
| *** geaaru has quit IRC | 18:28 | |
| timburke | tdasilva: dunno how much interest you've got in swiftonfile these days, but a bit ago someone nerd-sniped me into https://github.com/tipabu/swiftonfile/commit/7a6ba11 | 18:28 |
| *** saint_ has joined #openstack-swift | 18:37 | |
| *** bigdogstl has joined #openstack-swift | 18:43 | |
| *** armaan_ has quit IRC | 18:44 | |
| *** bigdogstl has quit IRC | 18:47 | |
| *** bigdogstl has joined #openstack-swift | 18:50 | |
| *** mhks has left #openstack-swift | 18:50 | |
| *** bigdogstl has quit IRC | 18:56 | |
| *** bigdogstl has joined #openstack-swift | 19:01 | |
| *** armaan has joined #openstack-swift | 19:06 | |
| *** bigdogstl has quit IRC | 19:13 | |
| *** itlinux has quit IRC | 19:20 | |
| *** bigdogstl has joined #openstack-swift | 19:21 | |
| *** saint_ has quit IRC | 19:21 | |
| openstackgerrit | Merged openstack/swift master: Add Docstrings to validate_replicas_by_tier https://review.openstack.org/535675 | 19:23 |
| *** bigdogstl has quit IRC | 19:25 | |
| *** itlinux has joined #openstack-swift | 19:28 | |
| *** bigdogstl has joined #openstack-swift | 19:58 | |
| *** bigdogstl has quit IRC | 20:03 | |
| clayg | timburke: on lp bug #1744375 - you're describing doing this GET request for the segments in order *before writing* the manifest? So that you can PUT the manifest down with this validated whole entire md5 value? | 20:11 |
| openstack | Launchpad bug 1744375 in OpenStack Object Storage (swift) "Include original object md5sum in sysmeta for better record keeping with SLOs" [Wishlist,Confirmed] https://launchpad.net/bugs/1744375 | 20:11 |
| clayg | do you then put that into the container? | 20:11 |
| clayg | maybe this is a verified or checked *LO - SLO 2.0? | 20:11 |
| timburke | well, i was thinking write-the-manifest-then-verify, but yeah. and for the truly paranoid, on read, check it again. won't be able to get the client an error, but you can at least log a warning | 20:13 |
| clayg | write-the-manifest-then-overwrite-it-with-sysmeta (?) | 20:13 |
| timburke | write the manifest with (unverified) sysmeta | 20:14 |
| timburke | and include the verification result in the final response | 20:14 |
| clayg | oh... so the client would PUT with "unverified-sysmeta: <hash>" and then we can 5XX in the end if there is a problem (but it's already written!?) | 20:15 |
| clayg | I don't think it's great to have manifests in durable storage where the value is sometimes totally checked and valid and sometimes garbage | 20:15 |
| clayg | I like the idea of calculating the hash from a read and using heartbeat - that's brilliant | 20:15 |
| clayg | I think S3's manifest/finalize takes a little while for similar sorts of reasons? | 20:16 |
| timburke | then the PUT can succeed, client times out or something waiting on the verification, and we've still made progress | 20:16 |
| timburke | maybe it's worth doing the reads on finalize? *shrug* hadn't thought that far ahead | 20:16 |
| timburke | er, "as part of the verification process" | 20:17 |
| timburke | in place of the HEADs | 20:17 |
| timburke | i'm 99% certain that S3 does something like what we do, based on their etags | 20:17 |
| timburke | i want *better* | 20:18 |
| timburke | (part of why i'd thought about it the way i had was that i was thinking of a separate filter *in front of* StaticLargeObject -- that beast's becoming unwieldy... we could still wrap it up in the slo entrypoint, but break it up a little like we did for encrypter/decrypter) | 20:23 |
| timburke | come to think of it... that may be a useful thing to do with https://review.openstack.org/#/c/499260/ too... then we don't have to worry about people putting them in the wrong order, or what happens when you have tlo but no slo... | 20:26 |
| patchbot | patch 499260 - swift - WIP: Add TLO middleware | 20:26 |
| *** bigdogstl has joined #openstack-swift | 20:29 | |
| *** bigdogstl has quit IRC | 20:40 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: Add TLO middleware https://review.openstack.org/499260 | 20:50 |
| *** bigdogstl has joined #openstack-swift | 21:02 | |
| *** bigdogstl has quit IRC | 21:06 | |
| *** bigdogstl has joined #openstack-swift | 21:36 | |
| *** bigdogstl has quit IRC | 21:45 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Functional tests for staticweb middleware https://review.openstack.org/503262 | 21:45 |
| *** tesseract has quit IRC | 22:01 | |
| *** bigdogstl has joined #openstack-swift | 22:06 | |
| *** bigdogstl has quit IRC | 22:11 | |
| *** ukaynar has joined #openstack-swift | 22:17 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Clean up comment about 404s leaking out during COPYs https://review.openstack.org/535981 | 22:24 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Functional tests for staticweb middleware https://review.openstack.org/503262 | 22:25 |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: Add TLO middleware https://review.openstack.org/499260 | 22:27 |
| *** chsc has quit IRC | 22:35 | |
| *** bigdogstl has joined #openstack-swift | 22:37 | |
| timburke | i think i may be loving this new zuulv3 business :-) http://logs.openstack.org/62/503262/4/check/swift-tox-func-domain-remap-staticweb/98eac38/testr_results.html.gz | 22:39 |
| torgomatic | clayg: looking at 48d94d96, it's probably because I was screwing around with range stuff and saw errors. If you wanted to make 401 <= X < 500 into not-errors, that's cool. The only one I'd want to keep is 400 because that sometimes means you got goofed-up rings. | 22:41 |
| clayg | it *does* !? | 22:42 |
| torgomatic | if you put a container server in your object ring or something, it'll 400 at you | 22:42 |
| *** bigdogstl has quit IRC | 22:43 | |
| torgomatic | I've seen it happen a couple times in this channel when someone builds rings the same but uses the account port for everything | 22:43 |
| torgomatic | OTOH, there's probably nice easy ways for users to get 400s just by putting dumb stuff in their requests, so meh | 22:44 |
| torgomatic | I don't even know if those backend 400s would show up in statsd, TBH | 22:44 |
| *** ukaynar has quit IRC | 22:47 | |
| *** ukaynar has joined #openstack-swift | 22:47 | |
| *** ukaynar has quit IRC | 22:47 | |
| *** ukaynar has joined #openstack-swift | 22:48 | |
| *** ukaynar_ has joined #openstack-swift | 22:49 | |
| *** ukaynar has quit IRC | 22:49 | |
| *** ukaynar_ has quit IRC | 22:49 | |
| *** ukaynar has joined #openstack-swift | 22:49 | |
| timburke | is *that* why we return stupid 412s when the client screws up?? | 22:53 |
| timburke | now i kinda wanna send x-backend headers to say "hey, you're an *object server*, right?" and have the server reject early based on that | 22:56 |
| timburke | maybe even use Expect: and 417? | 22:56 |
| *** itlinux has quit IRC | 23:00 | |
| *** itlinux has joined #openstack-swift | 23:02 | |
| openstackgerrit | Tim Burke proposed openstack/swift feature/deep: Merge remote-tracking branch 'origin/master' into feature/deep https://review.openstack.org/535990 | 23:04 |
| *** bigdogstl has joined #openstack-swift | 23:05 | |
| *** geaaru has joined #openstack-swift | 23:15 | |
| *** bigdogstl has quit IRC | 23:16 | |
| *** bigdogstl has joined #openstack-swift | 23:20 | |
| *** bigdogstl has quit IRC | 23:27 | |
| *** bigdogstl has joined #openstack-swift | 23:27 | |
| *** ukaynar has quit IRC | 23:37 | |
| *** bigdogstl has quit IRC | 23:40 | |
| -openstackstatus- NOTICE: Zuul will be offline over the next 20 minutes to perform maintenance; active changes will be reenqueued once work completes, but new patch sets or approvals during that timeframe may need to be rechecked or reapplied as appropriate | 23:43 | |
| *** corvus is now known as mr_book | 23:45 | |
| *** mr_book is now known as corvus | 23:45 | |
| *** itlinux has quit IRC | 23:47 | |
| *** bigdogstl has joined #openstack-swift | 23:49 | |
| *** itlinux has joined #openstack-swift | 23:50 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!