| *** haomaiwang has quit IRC | 00:01 | |
| *** haomaiwang has joined #openstack-swift | 00:01 | |
| kota_ | good morning | 00:03 |
|---|---|---|
| openstackgerrit | Paul Dardeau proposed openstack/swift: Remove limit of 64K devices in ring https://review.openstack.org/264922 | 00:05 |
| *** dmorita_ has quit IRC | 00:05 | |
| *** dmorita has joined #openstack-swift | 00:06 | |
| ho_ | kota_: morning! | 00:06 |
| *** dmorita has quit IRC | 00:06 | |
| *** dmorita has joined #openstack-swift | 00:06 | |
| kota_ | ho_: o/ | 00:07 |
| kota_ | ho_: btw, does your RBAC patch (patch 202411) include the direction to something like online docs for RBAC? | 00:08 |
| patchbot | kota_: https://review.openstack.org/#/c/202411/ - Add functional test for access control (RBAC) with... | 00:08 |
| *** asettle has quit IRC | 00:08 | |
| *** asettle has joined #openstack-swift | 00:09 | |
| kota_ | ho_: i wanna review but not sure for the expected behavior so i want some docs for double checking. | 00:09 |
| kota_ | ho_: i didn't start to review yet though :P | 00:09 |
| *** darrenc_afk is now known as darrenc | 00:12 | |
| *** zhill has quit IRC | 00:14 | |
| ho_ | kota_: thanks! there is a link in a comment (10:24am in Nov 9). if you click it, you can see some urls. i summarized expected result in the docs from code. | 00:16 |
| *** dmorita has quit IRC | 00:16 | |
| *** dmorita has joined #openstack-swift | 00:17 | |
| kota_ | ho_: in gerrit comment? | 00:17 |
| *** nadeem has quit IRC | 00:18 | |
| ho_ | kota_: yes. | 00:20 |
| *** asettle has quit IRC | 00:20 | |
| *** asettle has joined #openstack-swift | 00:20 | |
| kota_ | ho_: sorry, i found no http link there, i might be missing something... | 00:21 |
| *** dmorita has quit IRC | 00:21 | |
| *** dmorita has joined #openstack-swift | 00:21 | |
| kota_ | ho_: you commented, "@Jordan, thanks for the comment. I think it's not necessary to change proxy-server.conf for this patch and I put 6th user to sample.conf. After this patch will be lauded I will submit a patch to add users to run these tests." | 00:21 |
| kota_ | at 10:24am in Nov.9 | 00:21 |
| ho_ | kota_: one commen above. http://paste.openstack.org/show/478061/ | 00:22 |
| *** asettle has quit IRC | 00:22 | |
| ho_ | kota_: sorry two comments | 00:22 |
| *** asettle has joined #openstack-swift | 00:22 | |
| kota_ | ho_: all right | 00:22 |
| kota_ | ho_: looks like experimental result, right? no docs of keystone? | 00:23 |
| mattoliverau | kota_, ho_: morning | 00:23 |
| ho_ | kota_: yeah, (expected result from code) i think no doc for it. | 00:25 |
| ho_ | mattoliverau: morning! | 00:25 |
| *** gyee has quit IRC | 00:28 | |
| *** ho_away has joined #openstack-swift | 00:29 | |
| *** yarkot has joined #openstack-swift | 00:29 | |
| kota_ | ho_: ok, i think that matrix you made is already worthful :-) | 00:30 |
| *** mingdang1 has joined #openstack-swift | 00:30 | |
| *** gyee has joined #openstack-swift | 00:30 | |
| kota_ | ho_: i could take small time today for this and would be able to look at in detail next week (sorry, not so enough time for now) | 00:31 |
| ho_ | kota_: np! i really appreciate it :-) | 00:31 |
| *** mingdang1 has quit IRC | 00:38 | |
| *** achhabra has quit IRC | 00:39 | |
| *** m_kazuhiro has joined #openstack-swift | 00:41 | |
| *** dmorita has quit IRC | 00:42 | |
| *** dmorita has joined #openstack-swift | 00:43 | |
| *** lpabon has quit IRC | 00:44 | |
| *** dmorita has quit IRC | 00:47 | |
| *** ho_away has quit IRC | 00:48 | |
| clayg | gateway_role = self.node.org.noderole_set.first(runs_gateway=True) | 00:50 |
| clayg | hrmmm... | 00:50 |
| *** dmorita has joined #openstack-swift | 00:50 | |
| *** bwall has quit IRC | 00:53 | |
| *** bwall has joined #openstack-swift | 00:57 | |
| *** bwall has joined #openstack-swift | 00:57 | |
| *** jlvillal has quit IRC | 00:58 | |
| ho_ | clayg: could you please check my comment on patch 266190? (short summary is 2 replicas is no chance but 4 replicas is reasonable approach for the problem) | 01:00 |
| patchbot | ho_: https://review.openstack.org/#/c/266190/ - Fix posting accounts behavior when half of account... | 01:00 |
| *** haomaiwang has quit IRC | 01:01 | |
| ho_ | clayg: i will start to write unittest soon. | 01:01 |
| *** haomaiwa_ has joined #openstack-swift | 01:01 | |
| clayg | ho_: I'm not sure I understand your comment :) | 01:02 |
| *** jlvillal has joined #openstack-swift | 01:02 | |
| clayg | ho_: my comment was about the behavior with three replicas - but yeah the general idea is that the node that says 204 might not be "correct" - or worse - it could be only copy of the operation (so it'd be unfair to return 2XX to the client) | 01:02 |
| clayg | ho_: given that constrain for 3-replica, we can begin to consider different configurations - but we must preserve the symantic behavior of the api in the 3-replica case | 01:03 |
| clayg | the reason the 3-replica DELETE case can translate 404->2XX is because it writes the tombstone regardless. There's some risk that DELETE t1 gets (503, 404, 2XX) -> 2XX then DELETE t2 gets (204, 404, 503) -> 2XX - but FWIW everyone seemed to agree it was better than the alternatives we had at the time and despite giving the wrong answer (2XX a delete that should have 404'd) the risk seemed nominal compared to the undesireabl | 01:07 |
| ho_ | clayg: wait a minutes i'm trying understand... | 01:08 |
| clayg | in your case the 503 is still just as undeseriable - but the risk is higher - on the container/account layer 404 means nothing is stored, so we have nothing like a quorm - worse yet, it's possible the correct response would have 404 | 01:08 |
| ho_ | clayg: "in your case" means 4 replicas case? | 01:09 |
| clayg | ho_: no, i should have said in the account/container case - it's still the same for 3-replica | 01:09 |
| clayg | I think my comments here could have been more clear - the 404->204 transloation makes sense for the object layer DELETE only because of how the DELETE 404 case is handled | 01:11 |
| clayg | ... in the object-server's ObjectController | 01:11 |
| ho_ | clayg: i see. i understand less than 4 replicas we might loss data if a node which has a replica never comes back. | 01:12 |
| clayg | ho_: ok that's a fair rephrasing - so if the quorum value is 2/3 we have to have two successful opts to return 2xx - that's the deal ;) | 01:14 |
| clayg | ho_: but the point I was trying to make was that the "successfulness" of a 404 on an object server DELETE is *dramatically* different than the account/container 404 case for POST or DELETE | 01:15 |
| *** dmorita has quit IRC | 01:15 | |
| *** dmorita has joined #openstack-swift | 01:16 | |
| ho_ | clayg: yeah, I understand well (current three replica behavior is really well-designed). if we have 4 replicas and account/container 404 case for post or delete, we can gurantee the realiability (same level as 3 replica) when we replace a 404 to 204 | 01:17 |
| *** haomaiwa_ has quit IRC | 01:17 | |
| *** yarkot_ has joined #openstack-swift | 01:18 | |
| clayg | ho_: is that so? is (204, 204, 404, 503) really meeting the durability guarantee for 4 replicas? | 01:18 |
| *** dmorita has quit IRC | 01:19 | |
| *** dmorita has joined #openstack-swift | 01:19 | |
| clayg | ho_: what about (204, 404, 404, 503) - I believe the 404 won't quorum, so you'll move into translation and return success after having only one replica of the data? | 01:19 |
| clayg | ... i might be confused about when/how the realtive translations under make-request are applied | 01:20 |
| *** dmorita has quit IRC | 01:20 | |
| ho_ | clayg: my challenge is failed :-) my approach can't save second one | 01:20 |
| *** dmorita has joined #openstack-swift | 01:21 | |
| clayg | ho_: it's possible i'm mis-reading best_reponse - but I think that's how it works | 01:21 |
| ho_ | clayg: at least two 204 are necessary to gurantee data is my understanding so my current patch doesn't work for the case. | 01:21 |
| clayg | either way I think the point is more about treating a 404 from a storage server as a 204 - it's only appropriate to do so if the behavior of the storage server is know to be equivilent for those status codes - then I think the number of replicas matters not | 01:22 |
| clayg | ho_: well that's sort of orthagonal - but it's come up many times if the quorum of even replicas should be changed from >50% to >=50% - I'm not sure I disagree that "at least two [in a 4-replica ring] is sufficent to guarantee data" - but that's not how swift currently calculates quorum - today a 2xx response from a 4-replica policy should return at least 3 successful responses | 01:24 |
| ho_ | clayg: i understand but i would like to solve the problem :) | 01:25 |
| clayg | ho_: yeah I'm down! | 01:25 |
| clayg | ho_: which problem? specficially mult-region clusters returning bad responses? | 01:26 |
| ho_ | clayg: 2 region - 4 replicas and one region down | 01:26 |
| ho_ | clayg: deployers want to have same config in a region... | 01:28 |
| clayg | ho_: so fwiw my idea on metadata-layer updates for geo-clusters is to over-replicate and under-quorum. Basically allow an operator to explicitly craft a deployment where there's 3 account/container db's in each region and metadata updates across the region only happen through replication | 01:29 |
| *** yarkot_ has quit IRC | 01:29 | |
| clayg | ho_: it ties into sam's ideas on composite rings - region A has a container ring with *its* devices, the db's and updates go where they go (but only locally) - similar setup in region B | 01:30 |
| clayg | but then the container *replicator* has a compostie ring with all the replicas and syncs the updates out of band | 01:30 |
| ho_ | clayg: unnn. sounds great | 01:31 |
| clayg | ho_: fuck yeah - you should do it! | 01:31 |
| clayg | torgomatic can review it | 01:31 |
| *** dmorita has quit IRC | 01:31 | |
| * clayg drops the mic | 01:31 | |
| ho_ | clayg: i have to study a lot :-) | 01:31 |
| ho_ | clayg: thanks a lot! | 01:31 |
| clayg | ho_: I think it's very helpful just to raise the pain | 01:31 |
| *** nadeem has joined #openstack-swift | 01:32 | |
| clayg | notmyname: you see ho_ feeling the pain!? | 01:32 |
| *** dmorita has joined #openstack-swift | 01:33 | |
| clayg | ho_: but anyway - there might be better ideas | 01:36 |
| clayg | ho_: another idea maybe is to allow for "loose_quorum" on even replicas | 01:37 |
| clayg | ho_: another idea might be to udpate the account/container servers so they can record the request for metadata/delete operations even when the db does not exist | 01:37 |
| ho_ | clayg: actually my proposed patches are based on my pains (user troubles) | 01:37 |
| clayg | ho_: yes I understand, I am sorry the solution is not stright forward - but I don't see how we can preserve the durability guarentees we expect/require in the 3-replica case with those changes | 01:39 |
| clayg | ho_: but I'll be the first to admit multi-region metadata case is not satisfactory | 01:39 |
| clayg | well... rather not satisfactory *to me* | 01:39 |
| ho_ | clayg: i thought last idea but i would like to avoid cost for db copy | 01:39 |
| clayg | ho_: yup it's the sux - and might have negative consequences in the single region 3-replica case which could prevent it going in without some sort of flag/option | 01:40 |
| clayg | ho_: but better to not have flags/options and just be awesome all the time! | 01:41 |
| ho_ | clayg: i" metadata updates across the region only happen through replication" sounds best, other (my approach if it solves the problem) sounds like workaround... | 01:43 |
| *** nadeem has quit IRC | 01:44 | |
| clayg | ho_: yeah it's be pretty fucking sweet | 01:46 |
| ho_ | clayg: i will explain this discussion to users (for thier current problem). thank you very much! | 01:46 |
| clayg | bah - good luck :'( | 01:47 |
| clayg | sux | 01:47 |
| *** awelleck has joined #openstack-swift | 01:48 | |
| *** ho_away has joined #openstack-swift | 01:49 | |
| *** ho_away has quit IRC | 01:53 | |
| *** dkalleg has joined #openstack-swift | 01:53 | |
| dkalleg | Hi, I notice the swift client doesn't have a quota-set api call like some other services. Is there some other way to set swift quotas, or am I missing some fundamental swift concept? | 01:54 |
| timburke | dkalleg: you should be able to specify it with the --meta argument, like in `swift post container --meta 'quota-bytes: 1000000000'` | 01:57 |
| timburke | that also works for accounts | 01:58 |
| dkalleg | timburke: I see, thanks. | 02:00 |
| *** noark9 has joined #openstack-swift | 02:01 | |
| *** awelleck has quit IRC | 02:05 | |
| *** haomaiwang has joined #openstack-swift | 02:05 | |
| dkalleg | timburke: I'm trying to glean some info from here: http://docs.openstack.org/juno/config-reference/content/object-storage-account-quotas.html and from the swift help menu. I'm not seeing info on what some of these flags mean, -AUK | 02:12 |
| dkalleg | Is there a better set of docs somewhere I can refer to? | 02:12 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Follow up for Kazuhiro https://review.openstack.org/271119 | 02:13 |
| timburke | dkalleg: -A is for the auth url, -U the username, and -K the api key (password). those are all assuming v1 auth; keystone (v2 or v3) gets more complicated, but swift recognizes (to my knowledge) all the standard environment variables | 02:15 |
| dkalleg | ok, thats good, wasn't seeing a confirmation it would consume OS_xxx env vars. | 02:16 |
| dkalleg | I could end up asking a bunch more questions, but if you want to teach me how to fish instead of feeding me, I'd be happy to read up on it myself. Still just not finding docs that explain most of this stuff | 02:18 |
| *** mingdang1 has joined #openstack-swift | 02:18 | |
| timburke | dkalleg: swift --os-help has a nice long description of all the openstack auth options | 02:18 |
| *** asettle has quit IRC | 02:18 | |
| dkalleg | Or is docs.openstack.org the best there is? | 02:18 |
| dkalleg | ah ok, cool | 02:18 |
| *** asettle has joined #openstack-swift | 02:18 | |
| dkalleg | I didn't realize --os-help would give more | 02:18 |
| dkalleg | thanks | 02:18 |
| dkalleg | I | 02:19 |
| dkalleg | whoops, nvm | 02:19 |
| timburke | the usage strings are probably your best bet; http://docs.openstack.org/developer/python-swiftclient/ is rather api-heavy and cli-light | 02:19 |
| dkalleg | timburke: thanks, will give this a look | 02:20 |
| mingdang1 | swift upload a large file,swift can split the large file to little? | 02:29 |
| dkalleg | timburke: one more question, the docs seem to suggest that once you set quota-bytes for an account, any put with data greater than that quota would fail. So is this telling me the user can put an unlimited number of objects of the quota size or less? Or is that quota size a cumulative size for all of that users objects in storage? | 02:30 |
| *** ho_away has joined #openstack-swift | 02:32 | |
| *** asettle has quit IRC | 02:35 | |
| timburke | dkalleg: cumulative. though it should be noted that due to eventual consistency, users may exceed their quota in some situations (or be prevented from uploading new data until account updates propagate) | 02:35 |
| dkalleg | timburke: gotcha, thanks again | 02:36 |
| *** asettle has joined #openstack-swift | 02:37 | |
| *** haomaiwang has quit IRC | 03:01 | |
| *** 77CAAARUP has joined #openstack-swift | 03:01 | |
| *** sanchitmalhotra has joined #openstack-swift | 03:06 | |
| *** ho_away has quit IRC | 03:22 | |
| *** links has joined #openstack-swift | 03:26 | |
| jrichli | acoles tdasilva: back to slo and versioning, I am wondering what container(s) we expect the user to version. the normal container, the segment container, or both? timburke, have you ever used slo with versioning? | 03:27 |
| *** wasmum- has quit IRC | 03:33 | |
| *** wasmum has joined #openstack-swift | 03:36 | |
| jrichli | or maybe if you are doing slo with versioning, you shouldn't use a different container for segments | 03:37 |
| *** venkat has joined #openstack-swift | 03:38 | |
| openstackgerrit | Merged openstack/swift: Fix memory/socket leak in proxy on truncated SLO/DLO GET https://review.openstack.org/270234 | 03:39 |
| *** links has quit IRC | 03:40 | |
| openstackgerrit | Merged openstack/swift: Fix memory/socket leak in proxy on truncated SLO/DLO GET https://review.openstack.org/270233 | 03:41 |
| openstackgerrit | Kazuhiro MIYAHARA proposed openstack/swift: Fix handling of "Permission Denied" error from NamedTemporaryFile function https://review.openstack.org/270607 | 03:44 |
| *** asettle is now known as asettle-afk | 03:45 | |
| *** bill_az has quit IRC | 03:46 | |
| *** links has joined #openstack-swift | 03:48 | |
| notmyname | hello, world | 03:49 |
| openstackgerrit | Kazuhiro MIYAHARA proposed openstack/swift: Fix handling of "Permission Denied" error from NamedTemporaryFile function https://review.openstack.org/270607 | 03:49 |
| notmyname | just before lunch I got pulled into a all-day meeting. looks like I missed a bunch | 03:49 |
| jrichli | did you accomplish great things in the meeting, though? | 03:52 |
| notmyname | ...including peluse_ leaving. looks like clayg already made the "maybe if I -2 this, he'll stay" joke | 03:53 |
| *** ho_away has joined #openstack-swift | 03:54 | |
| notmyname | jrichli: oh, I hope so. mostly arguing, though. it was design-related stuff for some thing we're working on at swiftstack | 03:54 |
| notmyname | (I mean that in a friendly way) | 03:54 |
| notmyname | it was good | 03:54 |
| jrichli | i figured that is the type of arguing you meant - healthy debate! | 03:54 |
| notmyname | doesn't look like the gate situation has improved any. still 32+ hours long | 04:00 |
| *** 77CAAARUP has quit IRC | 04:01 | |
| notmyname | a few things landed, but still several more in the queue before the release can be tagged. maybe it will clear out this weekend | 04:01 |
| openstackgerrit | Michael Barton proposed openstack/swift: go: multipart writer that matches swift https://review.openstack.org/271131 | 04:01 |
| *** haomaiwa_ has joined #openstack-swift | 04:01 | |
| *** links has quit IRC | 04:04 | |
| *** mrmoje has quit IRC | 04:09 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 04:09 |
| *** ho_away has quit IRC | 04:10 | |
| *** daemontool_ has quit IRC | 04:12 | |
| *** daemontool_ has joined #openstack-swift | 04:13 | |
| *** ho_away has joined #openstack-swift | 04:16 | |
| *** links has joined #openstack-swift | 04:16 | |
| *** daemontool_ has quit IRC | 04:18 | |
| *** links has quit IRC | 04:27 | |
| *** links has joined #openstack-swift | 04:27 | |
| openstackgerrit | Takashi Kajinami proposed openstack/swift: Fix proxy-server's support for chunked transferring in GET object https://review.openstack.org/256201 | 04:33 |
| openstackgerrit | Merged openstack/swift: Bump eventlet min version to 0.17.4 https://review.openstack.org/269811 | 04:39 |
| *** wanghua has quit IRC | 04:44 | |
| *** links has quit IRC | 04:50 | |
| briancline | omfsm @ the gate delay | 04:51 |
| *** ho_away has quit IRC | 04:53 | |
| jrichli | acoles tdasilva timburke: I think I have answered my own question - the segment containers should specify a 'X-Versions-Location'. each one could be different. | 04:54 |
| *** links has joined #openstack-swift | 04:57 | |
| *** haomaiwa_ has quit IRC | 05:01 | |
| *** esker has joined #openstack-swift | 05:01 | |
| *** haomaiwang has joined #openstack-swift | 05:01 | |
| mahatic_ | peluse_: damn! | 05:04 |
| *** links has quit IRC | 05:13 | |
| *** ppai has joined #openstack-swift | 05:13 | |
| *** haomaiwang has quit IRC | 05:13 | |
| *** dmorita has quit IRC | 05:19 | |
| *** links has joined #openstack-swift | 05:21 | |
| *** klrmn has quit IRC | 05:24 | |
| *** trifon has joined #openstack-swift | 05:31 | |
| *** esker has quit IRC | 05:32 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Add unit tests for checking responses in AccountController https://review.openstack.org/271142 | 05:34 |
| *** links has quit IRC | 05:36 | |
| *** links has joined #openstack-swift | 05:37 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:40 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Add unit tests for checking responses in AccountController https://review.openstack.org/271142 | 05:44 |
| *** tdasilva has quit IRC | 05:45 | |
| openstackgerrit | Merged openstack/swift: authors and changelog updates for 2.6.0 https://review.openstack.org/269911 | 05:49 |
| openstackgerrit | Merged openstack/python-swiftclient: Use bulk-delete middleware when available https://review.openstack.org/190887 | 05:49 |
| openstackgerrit | Merged openstack/swift: Validate X-Timestamps https://review.openstack.org/265488 | 05:50 |
| briancline | \o/ | 05:53 |
| *** links has quit IRC | 05:59 | |
| *** links has joined #openstack-swift | 06:01 | |
| *** tdasilva has joined #openstack-swift | 06:04 | |
| *** asettle-afk is now known as asettle | 06:18 | |
| *** asettle has quit IRC | 06:20 | |
| *** asettle has joined #openstack-swift | 06:20 | |
| *** links has quit IRC | 06:22 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Fix posting accounts behavior when half of account servers downed https://review.openstack.org/266190 | 06:22 |
| *** links has joined #openstack-swift | 06:23 | |
| ho_ | briancline: hello | 06:25 |
| *** ChubYann has quit IRC | 06:26 | |
| *** venkat has quit IRC | 06:35 | |
| *** links has quit IRC | 06:46 | |
| *** venkat has joined #openstack-swift | 06:47 | |
| *** links has joined #openstack-swift | 06:50 | |
| *** asettle has quit IRC | 06:50 | |
| *** silor has joined #openstack-swift | 07:02 | |
| *** mrmoje has joined #openstack-swift | 07:04 | |
| *** silor has quit IRC | 07:07 | |
| *** silor has joined #openstack-swift | 07:08 | |
| mahatic_ | acoles_: thanks for the heads up on master merge | 07:08 |
| *** links has quit IRC | 07:09 | |
| *** links has joined #openstack-swift | 07:09 | |
| *** mrmoje has quit IRC | 07:10 | |
| *** silor1 has joined #openstack-swift | 07:14 | |
| *** silor has quit IRC | 07:16 | |
| *** silor1 is now known as silor | 07:16 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Add unit tests for checking responses in ContainerController https://review.openstack.org/271160 | 07:17 |
| *** dmorita has joined #openstack-swift | 07:20 | |
| *** dmorita has quit IRC | 07:25 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Add unit tests for checking responses in AccountController https://review.openstack.org/271142 | 07:28 |
| zaitcev | Hmm. Looks like my hackathon invitation might have expired. | 07:30 |
| *** rcernin has joined #openstack-swift | 07:31 | |
| *** links has quit IRC | 07:31 | |
| *** zaitcev has quit IRC | 07:35 | |
| onovy | ho_: hi | 07:40 |
| onovy | ho_: just tried "your" bug in our production. r=4, geocluster, 2 regions, 2 repl / region. stopped one whole region, a+c+o | 07:41 |
| onovy | and guess what? everything worked fine | 07:41 |
| onovy | put for containers/objects is working | 07:41 |
| *** links has joined #openstack-swift | 07:48 | |
| *** ChanServ sets mode: +v cschwede | 07:52 | |
| *** links has quit IRC | 07:54 | |
| ho_ | onovy: really? nice to hear it. do you have proxy log for the request? i want to see log like (200, 200, 503, 503) then 503 | 07:55 |
| ho_ | onovy: btw there is no problem on object. | 07:55 |
| *** venkat has quit IRC | 07:57 | |
| ho_ | onovy: if you have log for post container/account, i would like to compare with expected result (https://review.openstack.org/#/c/266190/3/test/unit/proxy/controllers/test_account.py) | 07:58 |
| *** rledisez has joined #openstack-swift | 08:07 | |
| *** venkat has joined #openstack-swift | 08:09 | |
| *** geaaru has joined #openstack-swift | 08:11 | |
| *** daemontool has joined #openstack-swift | 08:18 | |
| *** links has joined #openstack-swift | 08:22 | |
| *** jith_ has joined #openstack-swift | 08:25 | |
| jith_ | hi all what is the best way to monitor the swift cluster.. i m trying to configure a statsd server with swift-recon... but stuck in between.. can anyone please provide a guiding url ? | 08:26 |
| *** links has quit IRC | 08:35 | |
| *** venkat has quit IRC | 08:38 | |
| *** arnox has joined #openstack-swift | 08:39 | |
| onovy | ho_: i will download logs | 08:40 |
| *** baojg has joined #openstack-swift | 08:44 | |
| onovy | ho_: don't have this lines in log at all. only Jan 22 07:30:15 sdn-swift-proxy1 swift-proxy-server: ERROR with Container server <hostname> Trying to HEAD /v1/AUTH_d_25/d_15120128: Connection refused (txn: tx644b8ee1b96143658496d-0056a1cc77) | 08:45 |
| *** jmccarthy has quit IRC | 08:45 | |
| *** jmccarthy has joined #openstack-swift | 08:45 | |
| *** daemontool has quit IRC | 08:48 | |
| onovy | ho_: i think reason is simple: proxy are using handoff location for a/c when primary fails | 08:48 |
| *** daemontool has joined #openstack-swift | 08:48 | |
| onovy | so i have (200, 200, 200, 200) i think | 08:48 |
| onovy | and 2 of them are handoffs | 08:48 |
| *** venkat has joined #openstack-swift | 08:52 | |
| *** acoles_ is now known as acoles | 09:00 | |
| *** daemontool_ has joined #openstack-swift | 09:03 | |
| *** jordanP has joined #openstack-swift | 09:04 | |
| ho_ | onovy: thanks for the info. i think there is no contaier db so it will return 404. https://github.com/openstack/swift/blob/master/swift/container/server.py#L552 | 09:05 |
| *** daemontool has quit IRC | 09:06 | |
| ho_ | onovy: but result is fine for you :-) | 09:07 |
| *** mrmoje has joined #openstack-swift | 09:09 | |
| acoles | good morning | 09:12 |
| ho_ | acoles: good morning! | 09:12 |
| mahatic_ | acoles: good morning! | 09:14 |
| janonymous | morning all | 09:15 |
| *** jistr has joined #openstack-swift | 09:15 | |
| *** mrmoje has quit IRC | 09:32 | |
| *** joeljwright has joined #openstack-swift | 09:33 | |
| *** ChanServ sets mode: +v joeljwright | 09:33 | |
| *** mrmoje has joined #openstack-swift | 09:33 | |
| acoles | peluse_: https://www.youtube.com/watch?v=NuEWR-Ur0SU | 09:38 |
| *** venkat has quit IRC | 09:38 | |
| ho_ | onovy: post container works if you executes put container then post container to created container during region down | 09:40 |
| *** mrmoje has quit IRC | 09:40 | |
| ho_ | onovy: if there is a container before region down. post contaienr to the contaier during region down should be failed with 503 (204, 204, 404, 404) | 09:42 |
| *** venkat has joined #openstack-swift | 09:50 | |
| *** arnox has left #openstack-swift | 09:52 | |
| *** baojg has quit IRC | 09:52 | |
| *** arnox has joined #openstack-swift | 09:52 | |
| *** baojg has joined #openstack-swift | 09:54 | |
| onovy | ho_: ah! can you talk about it with my colleage "kaleta"? | 09:55 |
| onovy | marek will try to reproduce it inside our lab and can work on fix (if it's needed) with you | 09:55 |
| onovy | kaleta: ^^ | 09:55 |
| onovy | ho_: in this table: http://paste.openstack.org/show/484089/ you have: container+PUT => 503 | 09:57 |
| onovy | we created containers fine where half servers were downed | 09:57 |
| *** hyakuhei_ has joined #openstack-swift | 09:58 | |
| *** haomaiwang has joined #openstack-swift | 09:59 | |
| kaleta | ho_: hi. If I understand you correctly, the 503 should occur with container POST when half servers down to containers created just before shutting down the servers. Am I right? | 10:00 |
| *** haomaiwang has quit IRC | 10:01 | |
| *** haomaiwa_ has joined #openstack-swift | 10:01 | |
| ho_ | onovy: yep. detailed condition to reproduce put container failure is really rare (replica of account is moving from handoff to the target node) so it's expected with your situation | 10:02 |
| kaleta | thanks | 10:05 |
| ho_ | kaleta: you are right. above my commen is for put container not post container. sorry for confusion. | 10:10 |
| *** mrmoje has joined #openstack-swift | 10:17 | |
| *** aix has joined #openstack-swift | 10:23 | |
| *** mrmoje has quit IRC | 10:30 | |
| *** haomaiwa_ has quit IRC | 10:32 | |
| *** esker has joined #openstack-swift | 10:33 | |
| *** esker has quit IRC | 10:38 | |
| openstackgerrit | Merged openstack/swift: Stop overriding builtin range https://review.openstack.org/270549 | 10:40 |
| *** mingdang1 has quit IRC | 10:41 | |
| *** hyakuhei_ has quit IRC | 10:44 | |
| *** venkat has quit IRC | 10:48 | |
| *** ho_ has quit IRC | 10:48 | |
| *** baojg has quit IRC | 10:58 | |
| *** venkat has joined #openstack-swift | 11:02 | |
| *** arnox has quit IRC | 11:03 | |
| *** arnox has joined #openstack-swift | 11:04 | |
| *** arnox has quit IRC | 11:11 | |
| *** arnox has joined #openstack-swift | 11:11 | |
| *** arnox has quit IRC | 11:12 | |
| *** arnox has joined #openstack-swift | 11:12 | |
| *** arnox has left #openstack-swift | 11:12 | |
| *** jith_ has quit IRC | 11:20 | |
| *** aix has quit IRC | 11:23 | |
| *** fthiagogv has joined #openstack-swift | 11:26 | |
| *** sanchitmalhotra has quit IRC | 11:34 | |
| *** noark9 has quit IRC | 11:35 | |
| *** kei_yama has quit IRC | 11:36 | |
| *** aix has joined #openstack-swift | 11:37 | |
| *** m_kazuhiro has quit IRC | 11:41 | |
| *** mingdang1 has joined #openstack-swift | 11:44 | |
| *** ChanServ sets mode: +v tdasilva | 11:47 | |
| *** openstackgerrit has quit IRC | 11:47 | |
| *** openstackgerrit has joined #openstack-swift | 11:47 | |
| *** venkat has quit IRC | 11:50 | |
| *** haomaiwang has joined #openstack-swift | 11:54 | |
| *** haomaiwang has quit IRC | 11:58 | |
| *** venkat has joined #openstack-swift | 12:01 | |
| *** SkyRocknRoll has quit IRC | 12:03 | |
| *** km has quit IRC | 12:11 | |
| *** hseipp has joined #openstack-swift | 12:15 | |
| *** hseipp has quit IRC | 12:15 | |
| *** hseipp has joined #openstack-swift | 12:15 | |
| *** hyakuhei_ has joined #openstack-swift | 12:19 | |
| *** dmorita has joined #openstack-swift | 12:19 | |
| openstackgerrit | renminmin proposed openstack/python-swiftclient: Add pre-check quota before object uploading https://review.openstack.org/260887 | 12:22 |
| *** dmorita has quit IRC | 12:24 | |
| *** mingdang1 has quit IRC | 12:24 | |
| *** openstackgerrit has quit IRC | 12:33 | |
| *** openstackgerrit has joined #openstack-swift | 12:33 | |
| *** proteusguy has quit IRC | 12:38 | |
| *** jordanP has quit IRC | 12:40 | |
| acoles | tdasilva: jrichli: i logged a bug re the SLO versioning issue we were discussing yesterday: bug 1537042 | 12:43 |
| openstack | bug 1537042 in OpenStack Object Storage (swift) "SLO manifests become the large object when versioned" [Undecided,New] https://launchpad.net/bugs/1537042 | 12:43 |
| acoles | if people think its not a bug but desired behaviour then we can debate it there ! | 12:43 |
| tdasilva | acoles: thanks, looking it up now | 12:49 |
| acoles | tdasilva: https://gist.github.com/alistairncoles/32a3f6674abfa1cbd6e5 <- func test | 12:50 |
| *** proteusguy has joined #openstack-swift | 12:51 | |
| acoles | tdasilva: i want to try that test on your patch | 12:51 |
| acoles | tdasilva: it fails on master | 12:51 |
| tdasilva | mmm...looking | 12:51 |
| *** haomaiwang has joined #openstack-swift | 12:55 | |
| *** ppai has quit IRC | 12:58 | |
| *** bill_az has joined #openstack-swift | 12:59 | |
| *** silor1 has joined #openstack-swift | 12:59 | |
| *** haomaiwang has quit IRC | 13:00 | |
| *** silor has quit IRC | 13:02 | |
| *** silor1 is now known as silor | 13:02 | |
| *** daemontool_ has quit IRC | 13:04 | |
| *** silor1 has joined #openstack-swift | 13:05 | |
| *** silor has quit IRC | 13:06 | |
| *** silor1 is now known as silor | 13:06 | |
| *** marcusrafael has quit IRC | 13:10 | |
| *** esker has joined #openstack-swift | 13:15 | |
| *** hyakuhei_ has quit IRC | 13:16 | |
| openstackgerrit | Merged openstack/python-swiftclient: Convert http response(byte string) to string in python3. https://review.openstack.org/213640 | 13:16 |
| *** mrmoje has joined #openstack-swift | 13:18 | |
| *** esker has quit IRC | 13:20 | |
| *** ho_away has joined #openstack-swift | 13:22 | |
| *** silor1 has joined #openstack-swift | 13:26 | |
| *** silor1 has quit IRC | 13:28 | |
| *** silor has quit IRC | 13:28 | |
| *** silor has joined #openstack-swift | 13:29 | |
| tdasilva | acoles: good description with the bug report and the test also looks good | 13:32 |
| *** diogogmt has quit IRC | 13:35 | |
| *** jordanP has joined #openstack-swift | 13:35 | |
| *** diogogmt has joined #openstack-swift | 13:36 | |
| acoles | tdasilva: thinking about this overnight...having versioned writes do its own 'copying' does help simplify things, makes it easier to reason about what should happen with an 'internal copy' vs client COPY | 13:37 |
| tdasilva | acoles: yeah, I like the idea of having more control | 13:38 |
| tdasilva | for example, i was really happy with the idea of reducing that extra HEAD request | 13:39 |
| acoles | tdasilva: yes! | 13:39 |
| acoles | tdasilva: of course it does mean that you cannot maintain the (buggy) master behavior with SLO's without having versioned_writes call copy hooks in SLO :/ which would defeat the purpose. So let's hope it is a bug and not deliberate. | 13:42 |
| *** venkat has quit IRC | 13:42 | |
| tdasilva | acoles: IMO it's a bug, but yeah, I think we need to get agreement from others. I think the fact that we have no cross-function testing between slo and versioning is also an argument for it being a bug | 13:43 |
| tdasilva | in fact, it's an area we need to work on, cross functionality testing | 13:43 |
| acoles | tdasilva: hehe - any feature not tested is by definition a bug :P | 13:44 |
| acoles | tdasilva: agree, i have also been thinking I'd like to see a func test job running with fast post enabled | 13:44 |
| jrichli | after the large obejct is copied, you would get an error retrieving that file if it was over limit, right? i would think that means its a bug | 13:45 |
| acoles | jrichli: idk, actually what happens if you do a client COPY of an SLO that is over 5GB? | 13:46 |
| acoles | I mean client initiated COPY request | 13:46 |
| *** diogogmt has quit IRC | 13:47 | |
| jrichli | oh, i guess i dont know. i was stuck on what had happened with dlo post-as-copy, but its prob diff with slo normal copy since that was the designed behavior | 13:48 |
| tdasilva | jrichli: I don't think you get an error on a GET if object is over 5GB | 13:50 |
| acoles | tdasilva: jrichli: how about a PUT to a versioning SLO manifest that has a Range header? Do we create a version object that is just the ranges from the original? :/ | 13:50 |
| tdasilva | the check is only on a PUT | 13:50 |
| acoles | my money is on yes | 13:50 |
| jrichli | tdasilva: ah, right, of course. again, that dlo post-as-copy had me all messed up | 13:51 |
| tdasilva | acoles, jrichli: sorry, i need to run out for a couple of hours, but will be back later | 13:51 |
| jrichli | have fun! | 13:52 |
| jrichli | acoles: thanks for writing a test. i am excited to look at it. I got as far as to think about how I would approach a test and to play around with some things | 13:52 |
| acoles | actually looks like the versioned_writes COPY does not include the Range header | 13:52 |
| acoles | jrichli: if you want to you could try to replicate a similar test for DLO (although the expected outcomes will differ) | 13:53 |
| acoles | I didn't look for a DLO+versioning test, so I guess you might want to look first | 13:54 |
| jrichli | acoles: sure, i can look into that. it will be over the weekend, probably. there is the test.functional.tests.TestObjectVersioning.test_versioning_dlo. but i expect there could be things we can add. | 13:55 |
| jrichli | so, you say a "versioning manifest". I thought versioning was dictated at a container level? | 13:56 |
| *** esker has joined #openstack-swift | 13:56 | |
| jrichli | so I was thinking for slo versioning, it would be about whether or not each segment was in a versioned container | 13:57 |
| *** mrmoje has quit IRC | 13:57 | |
| *** mrmoje has joined #openstack-swift | 13:57 | |
| jrichli | oh, so i guess that can still be right. you have only been talking about actions against the manifest. | 13:58 |
| jrichli | that would apply for requests against the segments | 13:59 |
| jrichli | i just wanted to be sure - if you have the manifest in a versioned container, that doesn't necessarily mean that the segments will be versioned. right? | 13:59 |
| *** lpabon has joined #openstack-swift | 14:02 | |
| *** esker has quit IRC | 14:03 | |
| *** mrmoje has quit IRC | 14:05 | |
| *** esker has joined #openstack-swift | 14:06 | |
| *** rledisez has quit IRC | 14:07 | |
| *** rledisez has joined #openstack-swift | 14:08 | |
| *** ho_away has quit IRC | 14:08 | |
| *** daemontool has joined #openstack-swift | 14:08 | |
| *** hyakuhei_ has joined #openstack-swift | 14:12 | |
| *** daemontool_ has joined #openstack-swift | 14:12 | |
| *** daemontool has quit IRC | 14:13 | |
| *** haomaiwang has joined #openstack-swift | 14:14 | |
| *** haomaiwang has quit IRC | 14:18 | |
| *** esker has quit IRC | 14:31 | |
| *** haomaiwang has joined #openstack-swift | 14:35 | |
| *** daemontool_ has quit IRC | 14:37 | |
| *** lcurtis has joined #openstack-swift | 14:38 | |
| *** haomaiwang has quit IRC | 14:39 | |
| *** rledisez has quit IRC | 14:40 | |
| *** mrmoje has joined #openstack-swift | 14:42 | |
| acoles | jrichli: right. | 14:47 |
| acoles | jrichli: and when I (sloppily) say "versioning manifest" I mean a manifest in a container with X-versions-location, that will therefore experience the versioning phenomenon. | 14:47 |
| *** pcaruana has joined #openstack-swift | 14:50 | |
| *** haomaiwang has joined #openstack-swift | 14:56 | |
| *** blmartin has joined #openstack-swift | 14:59 | |
| *** haomaiwang has quit IRC | 15:00 | |
| *** esker has joined #openstack-swift | 15:04 | |
| *** SkyRocknRoll has joined #openstack-swift | 15:11 | |
| *** mrmoje has quit IRC | 15:11 | |
| *** diogogmt has joined #openstack-swift | 15:12 | |
| *** haomaiwang has joined #openstack-swift | 15:17 | |
| *** jistr has quit IRC | 15:18 | |
| *** haomaiwang has quit IRC | 15:22 | |
| *** mrmoje has joined #openstack-swift | 15:30 | |
| *** mragupat has joined #openstack-swift | 15:30 | |
| *** arch-nemesis has joined #openstack-swift | 15:31 | |
| *** cebruns_ has quit IRC | 15:32 | |
| *** cebruns has joined #openstack-swift | 15:33 | |
| *** haomaiwang has joined #openstack-swift | 15:38 | |
| *** mrmoje has quit IRC | 15:38 | |
| *** rledisez has joined #openstack-swift | 15:40 | |
| *** arnox has joined #openstack-swift | 15:41 | |
| *** haomaiwang has quit IRC | 15:43 | |
| *** awelleck has joined #openstack-swift | 15:47 | |
| *** daemontool has joined #openstack-swift | 15:58 | |
| *** haomaiwang has joined #openstack-swift | 15:58 | |
| *** haomaiwang has quit IRC | 16:01 | |
| *** daemontool has quit IRC | 16:03 | |
| *** klrmn has joined #openstack-swift | 16:04 | |
| openstackgerrit | Merged openstack/swift: remove unneeded duplicate dict keys https://review.openstack.org/270523 | 16:04 |
| *** mragupat has quit IRC | 16:12 | |
| ntt | Hi, I have a problem with swift and bulk upload. If I create a tar.gz from fedora (locale = UTF8) with filenames with characters like àèìòù, it works. If I create the tar from windows10, characters are wrong and bulk upload doesn't work. It happens the same if I create a file from windows and I try to transfer to linux (without using swift): ls -alh show my wrong filename. Someone can help me please? | 16:12 |
| *** mragupat has joined #openstack-swift | 16:15 | |
| *** rcernin has quit IRC | 16:15 | |
| *** diazjf has joined #openstack-swift | 16:15 | |
| *** diazjf1 has joined #openstack-swift | 16:17 | |
| *** trifon has quit IRC | 16:18 | |
| *** diazjf has quit IRC | 16:20 | |
| *** diazjf has joined #openstack-swift | 16:21 | |
| *** diazjf1 has quit IRC | 16:22 | |
| onovy | ntt: so you have problem with windows encoding, not swift and bulk upload, right? :) | 16:30 |
| *** minwoob has joined #openstack-swift | 16:31 | |
| onovy | acoles: https://review.openstack.org/#/c/270523/ // reverify is obsolete. just write "recheck" and new +A is not needed | 16:32 |
| onovy | recheck = run "check" and if +2+A then run "gate" | 16:32 |
| *** mrmoje has joined #openstack-swift | 16:34 | |
| *** awelleck has quit IRC | 16:35 | |
| *** arch-nemesis has quit IRC | 16:40 | |
| *** awelleck has joined #openstack-swift | 16:42 | |
| *** jistr has joined #openstack-swift | 16:43 | |
| *** arch-nemesis has joined #openstack-swift | 16:43 | |
| *** zaitcev has joined #openstack-swift | 16:45 | |
| *** ChanServ sets mode: +v zaitcev | 16:45 | |
| *** silor has quit IRC | 16:49 | |
| *** mrmoje has quit IRC | 16:52 | |
| *** mrmoje has joined #openstack-swift | 16:52 | |
| *** nadeem has joined #openstack-swift | 16:52 | |
| *** rcernin has joined #openstack-swift | 16:54 | |
| *** mrmoje has quit IRC | 16:57 | |
| *** dslevin has joined #openstack-swift | 16:57 | |
| *** esker has quit IRC | 16:59 | |
| *** hyakuhei_ has quit IRC | 17:00 | |
| *** zhill has joined #openstack-swift | 17:03 | |
| *** arnox has quit IRC | 17:07 | |
| *** jistr has quit IRC | 17:08 | |
| *** jistr has joined #openstack-swift | 17:09 | |
| *** rickyrem has joined #openstack-swift | 17:10 | |
| onovy | notmyname: hi. is there any release to not release new version of swift-bench? newest version is ~2 years old and there and new features in git. Thanks | 17:11 |
| onovy | /any release/any reason/ | 17:11 |
| *** jistr has quit IRC | 17:14 | |
| *** jistr has joined #openstack-swift | 17:17 | |
| *** rledisez has quit IRC | 17:17 | |
| rickyrem | Do any of you know of a guide to reset swift with no containers? | 17:21 |
| *** haomaiwa_ has joined #openstack-swift | 17:21 | |
| *** klrmn has quit IRC | 17:23 | |
| *** haomaiwa_ has quit IRC | 17:26 | |
| *** esker has joined #openstack-swift | 17:27 | |
| timburke | jrichli: it depends on how you use it. in the general case, i think you're right; the segments contain will need versioning as well, although until patch 214922 there won't be any protection against deletes. if you're always using swiftclient, it's default behavior will get in your way; on overwrites, it'll want to clean up the old segments by default. if you're diligent enough to always use --leave-segments (or are using the API and | 17:27 |
| timburke | always create your SwiftService with 'leave_segments': True), you could probably get away with not versioning the segments container at all; it bakes in the file's size and mtime (or the current time) when uploading, so collisions seem unlikely. it won't protect against an attacker, but it should be sufficient for a fair number of uses | 17:27 |
| *** gyee_ has joined #openstack-swift | 17:27 | |
| patchbot | timburke: https://review.openstack.org/#/c/214922/ - Add delete markers to versioned_writes middleware | 17:27 |
| *** nadeem has quit IRC | 17:32 | |
| *** nadeem has joined #openstack-swift | 17:32 | |
| *** fthiagogv has quit IRC | 17:33 | |
| *** dmorita has joined #openstack-swift | 17:33 | |
| *** dmorita has quit IRC | 17:33 | |
| *** dmorita has joined #openstack-swift | 17:34 | |
| *** joeljwright has quit IRC | 17:34 | |
| *** dmorita has quit IRC | 17:35 | |
| timburke | rickyrem: `swift delete --all` will walk an entire account and attempt to delete all objects and containers. alternatively, if you don't actually want the account any more (and the cluster has allow_account_management turned on), you can delete the account itself; subsequent requests will (iirc) return a 410 Gone until the account reaper has had time to clean up all the objects and containers | 17:35 |
| *** dmorita_ has joined #openstack-swift | 17:35 | |
| *** mrmoje has joined #openstack-swift | 17:36 | |
| rickyrem | timburke: I moved my drives arround and cleaned them out outside of Swift. | 17:37 |
| rickyrem | I rebuilt my rings, but the storage nodes are still aware of the old drives for some reason. | 17:38 |
| *** haomaiwang has joined #openstack-swift | 17:42 | |
| *** jordanP has quit IRC | 17:42 | |
| *** hyakuhei_ has joined #openstack-swift | 17:42 | |
| *** esker has quit IRC | 17:44 | |
| *** mrmoje has quit IRC | 17:45 | |
| *** haomaiwang has quit IRC | 17:47 | |
| timburke | were these just the drives from the object ring(s), or the account and container rings as well? and were these all pulled and cleaned at once, or over some period of time? i'd expect that either (1) this was done gradually, so from swift's perspective it looked like a bunch of drive failures and replacements | 17:48 |
| timburke | or (2) this was done immediately, but only for the object ring(s), so there are container dbs out there that think some objects exist when they don't. and assuming no tombstone files were written as part of the cleanup, swift's processes (container-updater, maybe? i forget) won't know to fix them | 17:48 |
| rickyrem | I rebuilt the account, object, and container rings at the same time. | 17:50 |
| rickyrem | The drives that are now part of those new and old rings were rm -rf'ed | 17:50 |
| -openstackstatus- NOTICE: Restarting zuul due to a memory leak | 17:51 | |
| rickyrem | Is there something else I can delete to make Swift act like this is it's first time running? | 17:52 |
| *** esker has joined #openstack-swift | 17:52 | |
| *** thumpba has joined #openstack-swift | 17:54 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Include response headers in ClientExceptions https://review.openstack.org/269382 | 17:56 |
| *** jistr has quit IRC | 17:57 | |
| notmyname | onovy: good question. I know I looked at it a while back, but for whatever reason I didn't follow through with a release. I'll look into it again, soon | 17:58 |
| *** mragupat has quit IRC | 17:58 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Accept gzip-encoded API responses https://review.openstack.org/184956 | 18:00 |
| *** 14WAAR75L has joined #openstack-swift | 18:03 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Prompt for missing password https://review.openstack.org/189387 | 18:03 |
| *** esker has quit IRC | 18:05 | |
| *** 14WAAR75L has quit IRC | 18:08 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Add --decode-content option https://review.openstack.org/184957 | 18:08 |
| *** hyakuhei_ has quit IRC | 18:11 | |
| *** hseipp has quit IRC | 18:12 | |
| *** klrmn has joined #openstack-swift | 18:16 | |
| *** haomaiwang has joined #openstack-swift | 18:24 | |
| *** haomaiwang has quit IRC | 18:29 | |
| *** nadeem has quit IRC | 18:33 | |
| tdasilva | acoles, jrichli: back..looking through your comments on the patch right now, thanks for the review | 18:34 |
| *** jordanP has joined #openstack-swift | 18:36 | |
| acoles | tdasilva: np. the EC response not having swift_x_timestamp took some tracking down. I thought I had a fix but some unit tests are now failing, I think because timburke said that time can't be negative anymore. | 18:38 |
| acoles | that'll teach me to rebase! | 18:38 |
| *** geaaru has quit IRC | 18:38 | |
| *** lpabon has quit IRC | 18:40 | |
| *** haomaiwang has joined #openstack-swift | 18:45 | |
| *** jordanP has quit IRC | 18:46 | |
| onovy | notmyname: thanks | 18:48 |
| tdasilva | acoles: hmm..interesting..so is using x-timestamp a possiblity? | 18:48 |
| acoles | tdasilva: it should work, i think i have a local problem with the unit test making the epoch into negative time float :/ my machine thinks it has DST, i'm confused! | 18:49 |
| *** haomaiwang has quit IRC | 18:49 | |
| acoles | tdasilva: gotta go, back on it next week, have a good weekend | 18:50 |
| *** acoles is now known as acoles_ | 18:51 | |
| *** ChubYann has joined #openstack-swift | 18:54 | |
| timburke | acoles_: i can't help but wonder why you wanted a negative timestamp...i was pushing to allow them originally, but clayg said no (paraphrasing) :( | 18:57 |
| *** PsionTheory has joined #openstack-swift | 19:01 | |
| notmyname | ah, cool. looks like just one more patch in the gate queue that is referenced in this release | 19:05 |
| notmyname | patch 205803 | 19:05 |
| patchbot | notmyname: https://review.openstack.org/#/c/205803/ - Container-Sync to iterate only over synced containers | 19:05 |
| notmyname | estimated ~1 hour more | 19:06 |
| *** haomaiwang has joined #openstack-swift | 19:06 | |
| *** haomaiwang has quit IRC | 19:11 | |
| clayg | timburke: I really just said "this makes my brain hurt - what's the use case" and then you changed to the patch get rid of a 500 and without making my brain hurt so we merged it | 19:11 |
| clayg | timburke: acoles_: if there's a use case for negative timestamps I'm all ears! | 19:11 |
| *** rcernin has quit IRC | 19:12 | |
| *** esker has joined #openstack-swift | 19:13 | |
| notmyname | torgomatic: your patch 252096 has 2 +2s and another +1 from a core. do you have any idea why it doesn't have a +A? | 19:14 |
| patchbot | notmyname: https://review.openstack.org/#/c/252096/ - Allow smaller segments in static large objects | 19:14 |
| notmyname | cschwede: ^ you were 2nd +2 | 19:15 |
| openstackgerrit | Paul Dardeau proposed openstack/swift: Remove limit of 64K devices https://review.openstack.org/264922 | 19:17 |
| *** mragupat has joined #openstack-swift | 19:19 | |
| openstackgerrit | Ondřej Nový proposed openstack/swift-bench: Bump eventlet min version to 0.17.4 https://review.openstack.org/271487 | 19:22 |
| *** chmouel_ is now known as chmouel | 19:25 | |
| *** haomaiwang has joined #openstack-swift | 19:27 | |
| *** haomaiwang has quit IRC | 19:32 | |
| *** bjkeller has joined #openstack-swift | 19:35 | |
| *** haomaiwang has joined #openstack-swift | 19:48 | |
| *** haomaiwang has quit IRC | 19:53 | |
| *** zhill has quit IRC | 19:53 | |
| *** zhill_ has joined #openstack-swift | 19:53 | |
| *** hyakuhei_ has joined #openstack-swift | 20:04 | |
| *** haomaiwang has joined #openstack-swift | 20:09 | |
| *** haomaiwang has quit IRC | 20:13 | |
| *** lpabon has joined #openstack-swift | 20:14 | |
| *** pcaruana has quit IRC | 20:14 | |
| *** esker has quit IRC | 20:16 | |
| *** esker has joined #openstack-swift | 20:17 | |
| *** haomaiwang has joined #openstack-swift | 20:30 | |
| *** haomaiwang has quit IRC | 20:34 | |
| *** hyakuhei_ has quit IRC | 20:46 | |
| *** acoles_ is now known as acoles | 20:52 | |
| acoles | timburke: clayg the result of this line https://github.com/openstack/swift/blob/master/swift/common/middleware/versioned_writes.py#L212-212 is negative if the last-modified time is <hour after the epoch *and* your machine timezone is not UTC. tdasilva's patch 260179 coincidentally changes a unit test to use 1 sec after the epoch. so it broke after rebase onto master on my machine. | 21:03 |
| patchbot | acoles: https://review.openstack.org/#/c/260179/ - decouple versioned writes from COPY | 21:03 |
| acoles | after an hour trying to understand python timezone handling and mktime() i just changed my timezone to UTC :) | 21:04 |
| timburke | time is hard :( | 21:04 |
| *** mrmoje has joined #openstack-swift | 21:05 | |
| acoles | i still don't really understand it :/ nor if there's a real world bug there if a timestamp is within hours of the epoch | 21:06 |
| acoles | 'it' being localtime handling | 21:06 |
| *** number80 is now known as hguemar | 21:06 | |
| *** hguemar is now known as number80 | 21:07 | |
| acoles | timburke: the good news is that in my timezone it is beer time :) | 21:09 |
| *** acoles is now known as acoles_ | 21:12 | |
| *** marcusvrn_ has quit IRC | 21:17 | |
| timburke | that certainly *seems* like a bug; we're getting a GMT last-modified header, parsing it, and replacing the tzinfo with our local tz | 21:20 |
| *** nadeem has joined #openstack-swift | 21:23 | |
| *** luksaur has quit IRC | 21:23 | |
| timburke | looks like we want calendar.timegm instead of time.mktime? | 21:26 |
| pdardeau | redbo: thx for having a look at patch 264922 | 21:26 |
| patchbot | pdardeau: https://review.openstack.org/#/c/264922/ - Remove limit of 64K devices | 21:26 |
| *** nadeem has quit IRC | 21:26 | |
| *** nadeem has joined #openstack-swift | 21:27 | |
| pdardeau | redbo: would you recommend changing both methods to standalone functions? | 21:27 |
| redbo | I think so. Since they don't actually touch anything in the class or instantiate it. | 21:31 |
| timburke | oh man, see how smart clayg is? versioned_writes would've been all kinds of sad if i'd allowed negative timestamps! everything before 1970 would've come out in reverse! | 21:38 |
| torgomatic | isn't that the plot of Back to the Future or something? | 21:42 |
| *** cebruns has quit IRC | 21:44 | |
| *** cebruns has joined #openstack-swift | 21:46 | |
| *** eranrom has joined #openstack-swift | 21:47 | |
| *** pcaruana has joined #openstack-swift | 21:48 | |
| *** dkalleg has left #openstack-swift | 21:50 | |
| *** rcernin has joined #openstack-swift | 21:52 | |
| *** cebruns has quit IRC | 21:53 | |
| *** haomaiwang has joined #openstack-swift | 21:53 | |
| *** cebruns has joined #openstack-swift | 21:55 | |
| *** awelleck1 has joined #openstack-swift | 21:56 | |
| *** haomaiwang has quit IRC | 21:57 | |
| *** awelleck has quit IRC | 21:58 | |
| *** mragupat has quit IRC | 21:59 | |
| *** nadeem has quit IRC | 22:00 | |
| *** hyakuhei_ has joined #openstack-swift | 22:01 | |
| pdardeau | redbo: i had one of those methods as classmethod and the other as staticmethod -- which is wonky | 22:02 |
| pdardeau | redbo: the reason for marking staticmethod was in hopes of making things more understandable from where they're called. | 22:03 |
| *** nadeem has joined #openstack-swift | 22:03 | |
| pdardeau | so that seeing RingData.create_part2dev_array might be more meaningful than just create_part2dev_array. or at least that was my thinking | 22:04 |
| jrichli | timburke: thanks for the good info! | 22:05 |
| openstackgerrit | Tim Burke proposed openstack/swift: Use calendar.timegm not time.mktime https://review.openstack.org/271542 | 22:13 |
| timburke | acoles_: ^^ | 22:13 |
| *** minwoob has quit IRC | 22:18 | |
| *** mrmoje has quit IRC | 22:18 | |
| *** blmartin has quit IRC | 22:20 | |
| *** haomaiwang has joined #openstack-swift | 22:33 | |
| *** haomaiwang has quit IRC | 22:38 | |
| *** PsionTheory has quit IRC | 22:38 | |
| clayg | timburke: point of fact - clayg is exceptionally dumb - which just happens to be a benifit in working on distributed systems because it tends to keep you from trying to be too clever for your own good | 22:40 |
| clayg | timburke: something about "If the implementation is easy to explain, it may be a good idea" | 22:41 |
| *** hyakuhei_ has quit IRC | 22:46 | |
| openstackgerrit | Merged openstack/swift: Container-Sync to iterate only over synced containers https://review.openstack.org/205803 | 22:49 |
| notmyname | yay! | 22:50 |
| notmyname | so I've asked for permission to do the 2.6.0 release | 22:50 |
| notmyname | https://review.openstack.org/#/c/271555/ | 22:50 |
| notmyname | so when that one lands, we'll have it | 22:51 |
| *** rcernin has quit IRC | 22:53 | |
| *** rickyrem has left #openstack-swift | 22:57 | |
| *** diazjf has quit IRC | 22:58 | |
| *** bjkeller has left #openstack-swift | 23:03 | |
| *** jamielennox is now known as jamielennox|away | 23:15 | |
| *** awelleck1 has quit IRC | 23:16 | |
| *** nadeem has quit IRC | 23:22 | |
| *** shakamunyi has quit IRC | 23:22 | |
| *** bill_az has quit IRC | 23:34 | |
| *** thumpba has quit IRC | 23:34 | |
| *** lcurtis has quit IRC | 23:45 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!