| *** CR7 has quit IRC | 00:04 | |
| *** tongli has joined #openstack-swift | 00:09 | |
| *** tongli has quit IRC | 00:14 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix typo in Deployment Guide and add some formatting https://review.openstack.org/223356 | 00:17 |
|---|---|---|
| *** minwoob has quit IRC | 00:20 | |
| *** kota_ has joined #openstack-swift | 00:26 | |
| *** ChanServ sets mode: +v kota_ | 00:26 | |
| kota_ | good morning | 00:26 |
| *** m_kazuhiro has joined #openstack-swift | 00:29 | |
| *** dmorita has joined #openstack-swift | 00:34 | |
| mattoliverau | kota_: morning | 00:35 |
| kota_ | mattoliverau: morning | 00:36 |
| kota_ | mattoliverau: so... I updated my patch for bug 1460920 | 00:37 |
| openstack | bug 1460920 in OpenStack Object Storage (swift) "Successful PUT object might be missing container update on EC" [High,In progress] https://launchpad.net/bugs/1460920 - Assigned to Kota Tsuyuzaki (tsuyuzaki-kota) | 00:37 |
| kota_ | mattoliverau: if you could have a time, could you review it? | 00:37 |
| kota_ | patch is here, https://review.openstack.org/#/c/186735/ | 00:38 |
| mattoliverau | kota_: I'll do my best too :) | 00:38 |
| kota_ | I changed proxy to reattach missing container updates into footers of successful nodes. | 00:38 |
| kota_ | mattoliverau: thanks ;) | 00:39 |
| mattoliverau | kota_: nice, yeah that's the best way.. I still have to look at the code, but do we do it for both EC and Repl? Cause in theory don't we have the same problem if the object ring has many replica's vs container replica count? | 00:40 |
| kota_ | mattoliverau: exactly | 00:41 |
| kota_ | but we don't have a way to put footers yet in Repl. | 00:41 |
| kota_ | so i expect it's more big patch and it will be hard to review. | 00:41 |
| mattoliverau | yeah, I think footers need to be refactored out, which I think it will be cause doesn't encrytion need it | 00:42 |
| kota_ | so IMO, the first land should be a fix for EC, and then, we will fix about Repl too. | 00:42 |
| mattoliverau | kota_: totally agree, just putting it out there :) | 00:42 |
| *** zhill has quit IRC | 00:42 | |
| kota_ | mattoliverau: thanks but nice thought and notation to me :) | 00:43 |
| *** garthb has quit IRC | 00:45 | |
| *** haomaiwang has joined #openstack-swift | 00:59 | |
| *** haomaiwang has quit IRC | 01:01 | |
| *** haomaiwang has joined #openstack-swift | 01:01 | |
| *** haomaiwang has quit IRC | 01:05 | |
| *** haomaiwang has joined #openstack-swift | 01:06 | |
| *** haomaiwang has quit IRC | 01:09 | |
| *** asettle has joined #openstack-swift | 01:19 | |
| openstackgerrit | Zack M. Davis proposed openstack/swift: given Python 3, use parse_headers instead of rfc822.Message https://review.openstack.org/203304 | 01:22 |
| *** haomaiwang has joined #openstack-swift | 01:46 | |
| *** haomaiwang has quit IRC | 02:01 | |
| *** haomaiwang has joined #openstack-swift | 02:01 | |
| *** asettle is now known as asettle-afk | 02:05 | |
| *** haomaiwang has quit IRC | 02:13 | |
| *** DericHorn-HP has quit IRC | 02:18 | |
| *** haomaiwang has joined #openstack-swift | 02:21 | |
| *** baojg has joined #openstack-swift | 02:27 | |
| *** DericHorn-HP has joined #openstack-swift | 02:27 | |
| *** tongli has joined #openstack-swift | 02:38 | |
| *** tongli has quit IRC | 02:43 | |
| *** DericHorn-HP has quit IRC | 02:57 | |
| *** haomaiwang has quit IRC | 03:01 | |
| *** haomaiwang has joined #openstack-swift | 03:01 | |
| *** gyee has quit IRC | 03:01 | |
| *** DericHorn-HP has joined #openstack-swift | 03:05 | |
| *** mahatic has joined #openstack-swift | 03:21 | |
| mahatic | good morning | 03:23 |
| *** asettle-afk is now known as asettle | 03:24 | |
| *** sudorandom has quit IRC | 03:28 | |
| *** CrackerJackMack has quit IRC | 03:28 | |
| *** jroll has quit IRC | 03:28 | |
| *** darrenc_ has joined #openstack-swift | 03:29 | |
| *** zaitcev has quit IRC | 03:29 | |
| *** zacksh has quit IRC | 03:29 | |
| *** mattoliverau has quit IRC | 03:29 | |
| *** darrenc has quit IRC | 03:29 | |
| *** mattoliverau has joined #openstack-swift | 03:29 | |
| *** CrackerJackMack has joined #openstack-swift | 03:30 | |
| *** sudorandom has joined #openstack-swift | 03:30 | |
| *** ChanServ sets mode: +v mattoliverau | 03:30 | |
| *** zacksh has joined #openstack-swift | 03:30 | |
| mattoliverau | mahatic: morning | 03:31 |
| *** jroll has joined #openstack-swift | 03:31 | |
| *** nakagawamsa has joined #openstack-swift | 03:35 | |
| *** bill_az has quit IRC | 03:35 | |
| mahatic | mattoliverau: o/ are you better now? | 03:35 |
| mattoliverau | Meh, but still working.. Just slowly :) | 03:36 |
| mahatic | :) | 03:36 |
| asettle | *cringe* mattoliverau I ended up with 38 comments on that patch, probably should do it myself hey? :p | 03:38 |
| *** sudorandom has quit IRC | 03:39 | |
| *** jroll has quit IRC | 03:41 | |
| *** zacksh has quit IRC | 03:41 | |
| *** CrackerJackMack has quit IRC | 03:41 | |
| *** mattoliverau has quit IRC | 03:41 | |
| *** mattoliverau has joined #openstack-swift | 03:42 | |
| *** zacksh has joined #openstack-swift | 03:43 | |
| *** david-lyle has joined #openstack-swift | 03:50 | |
| *** haomaiwang has quit IRC | 04:01 | |
| *** haomaiwang has joined #openstack-swift | 04:01 | |
| *** CrackerJackMack has joined #openstack-swift | 04:03 | |
| *** sudorandom has joined #openstack-swift | 04:03 | |
| *** jroll has joined #openstack-swift | 04:04 | |
| *** rohit_ has joined #openstack-swift | 04:05 | |
| rohit_ | core reviewers: May i request reviews and approvals please : https://review.openstack.org/#/c/216903/ and https://review.openstack.org/#/c/217309/. These belong to ceilometermiddleware.swift | 04:06 |
| *** kota_ has quit IRC | 04:07 | |
| *** wshao has joined #openstack-swift | 04:07 | |
| *** zhill has joined #openstack-swift | 04:08 | |
| *** wshao has quit IRC | 04:17 | |
| hrou | mahatic, quick question for you, what were your probe test results using patch 214438 ? | 04:20 |
| patchbot | hrou: https://review.openstack.org/#/c/214438/ | 04:20 |
| hrou | mahatic, on the trello page it states "probetests: errors=2, failures=1 (probably different with mahatic changes)" but I'm seeing 2 failures my self. | 04:21 |
| *** darrenc_ is now known as darrenc | 04:21 | |
| mahatic | hrou: before my changes, the probe failures were similar to what is mentioned in etherpad. error=2 failures=1 | 04:24 |
| hrou | mahatic, what about after ? : ) | 04:24 |
| mahatic | hrou: none :D | 04:24 |
| *** DericHorn-HP has quit IRC | 04:25 | |
| hrou | By your changes are you referring to 214438 ? | 04:25 |
| mahatic | hrou: do you have probe test changes? | 04:25 |
| mahatic | hrou: nope. patch 220897 | 04:26 |
| patchbot | mahatic: https://review.openstack.org/#/c/220897/ | 04:26 |
| hrou | Ah do you mean patch 220897 | 04:26 |
| patchbot | hrou: https://review.openstack.org/#/c/220897/ | 04:26 |
| hrou | ah right : ) | 04:26 |
| hrou | Nope, I don't, let me pull that down and give it a go. | 04:26 |
| hrou | Actually that should be the new head of the patch chain now. Thanks mahatic ! I'll try that out. | 04:28 |
| mahatic | hrou: yup, that's correct | 04:28 |
| *** DericHorn-HP has joined #openstack-swift | 04:28 | |
| hrou | mahatic, great, thanks again! | 04:28 |
| mahatic | np | 04:29 |
| *** oc714|2 has joined #openstack-swift | 04:35 | |
| *** DericHorn-HP has quit IRC | 04:44 | |
| *** cdelatte has quit IRC | 04:45 | |
| *** cdelatte has joined #openstack-swift | 04:46 | |
| *** DericHorn-HP has joined #openstack-swift | 04:48 | |
| openstackgerrit | Alexandra Settle proposed openstack/swift: Moving DLO functionality doc to middleware https://review.openstack.org/219991 | 04:50 |
| *** haomaiwang has quit IRC | 05:01 | |
| *** haomaiwang has joined #openstack-swift | 05:01 | |
| *** ppai has joined #openstack-swift | 05:04 | |
| *** baojg has quit IRC | 05:05 | |
| *** garthb has joined #openstack-swift | 05:06 | |
| *** baojg has joined #openstack-swift | 05:07 | |
| *** hezhiqiang has joined #openstack-swift | 05:12 | |
| *** kota_ has joined #openstack-swift | 05:15 | |
| *** ChanServ sets mode: +v kota_ | 05:15 | |
| *** flwang1 has quit IRC | 05:18 | |
| *** DericHorn-HP has quit IRC | 05:20 | |
| *** esker has joined #openstack-swift | 05:34 | |
| *** mahatic has quit IRC | 05:35 | |
| asettle | notmyname, this is for you later: https://review.openstack.org/#/c/219991/5 | 05:36 |
| *** baojg has quit IRC | 05:42 | |
| *** DericHorn-HP has joined #openstack-swift | 05:45 | |
| *** DericHorn-HP has quit IRC | 05:46 | |
| *** hrou has quit IRC | 05:50 | |
| *** baojg has joined #openstack-swift | 05:51 | |
| *** frank__ has joined #openstack-swift | 05:58 | |
| *** garthb has quit IRC | 05:58 | |
| *** frank__ has quit IRC | 05:58 | |
| *** frank__ has joined #openstack-swift | 05:59 | |
| *** haomaiwang has quit IRC | 06:01 | |
| *** frank__ is now known as fgemein_ | 06:01 | |
| *** haomaiwang has joined #openstack-swift | 06:01 | |
| *** fgemein_ is now known as panchovilla | 06:01 | |
| panchovilla | hello everyone | 06:02 |
| panchovilla | i have some questions regarding the swift builder files ( e.g. account.builder ) | 06:04 |
| panchovilla | is there anyone with some deeper insight into this topic | 06:05 |
| panchovilla | ? | 06:05 |
| * panchovilla is a sysadmin from Hamburg | 06:09 | |
| panchovilla | LIST | 06:11 |
| *** m_kazuhiro has quit IRC | 06:12 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Use small part power instead of seeds https://review.openstack.org/223422 | 06:12 |
| kota_ | panchovilla: what do you want to ask? perhaps, I might answer for you. | 06:15 |
| kota_ | because I'm reading ring builder code just now | 06:16 |
| panchovilla | Hi Thx! I've restored builder out of the *.ring.gz | 06:20 |
| panchovilla | Now I want to check the results | 06:20 |
| panchovilla | Maybe you can take a short look at my question here ? https://ask.openstack.org/en/question/81708/ring-builder-files-lost-swift/ | 06:21 |
| *** zhill has quit IRC | 06:27 | |
| *** SkyRocknRoll has joined #openstack-swift | 06:29 | |
| *** hezhiqia_ has joined #openstack-swift | 06:33 | |
| *** hezhiqiang has quit IRC | 06:34 | |
| *** hezhiqiang has joined #openstack-swift | 06:38 | |
| *** hezhiqia_ has quit IRC | 06:38 | |
| kota_ | panchovilla: I'm back. | 06:38 |
| *** wshao has joined #openstack-swift | 06:42 | |
| *** hezhiqiang has quit IRC | 06:43 | |
| *** hezhiqia_ has joined #openstack-swift | 06:44 | |
| *** m_kazuhiro has joined #openstack-swift | 06:46 | |
| panchovilla | builder._set_parts_wanted() | 06:52 |
| panchovilla | Traceback (most recent call last): | 06:52 |
| panchovilla | File "<stdin>", line 1, in <module> | 06:52 |
| panchovilla | File "/usr/local/lib/python2.7/dist-packages/swift-1.7.5-py2.7.egg/swift/common/ring/builder.py", line 456, in _set_parts_wanted | 06:52 |
| panchovilla | int(weight_of_one_part * dev['weight']) - dev['parts'] | 06:52 |
| panchovilla | KeyError: 'parts' | 06:52 |
| *** haomaiwang has quit IRC | 07:01 | |
| *** haomaiwang has joined #openstack-swift | 07:01 | |
| *** rledisez has joined #openstack-swift | 07:09 | |
| *** wshao has quit IRC | 07:10 | |
| *** hezhiqiang has joined #openstack-swift | 07:12 | |
| *** hezhiqia_ has quit IRC | 07:16 | |
| *** haomaiwang has quit IRC | 07:18 | |
| *** haomaiwang has joined #openstack-swift | 07:19 | |
| *** jordanP has joined #openstack-swift | 07:38 | |
| *** geaaru has joined #openstack-swift | 07:39 | |
| *** mahatic has joined #openstack-swift | 07:41 | |
| *** panchovilla has quit IRC | 07:49 | |
| *** hezhiqiang has quit IRC | 07:56 | |
| *** acoles_ is now known as acoles | 07:57 | |
| *** hezhiqiang has joined #openstack-swift | 07:58 | |
| mahatic | acoles: good morning! | 07:58 |
| *** m_kazuhiro has quit IRC | 07:59 | |
| acoles | mahatic: good morning | 08:00 |
| *** rohit_ has quit IRC | 08:00 | |
| *** haomaiwang has quit IRC | 08:01 | |
| acoles | rohit_: swift cores cannot approve ceilometer middleware reviews. at least, i can't. | 08:01 |
| *** haomaiwa_ has joined #openstack-swift | 08:01 | |
| *** chenhuayi has joined #openstack-swift | 08:07 | |
| *** mahatic has quit IRC | 08:14 | |
| *** jistr has joined #openstack-swift | 08:18 | |
| *** m_kazuhiro has joined #openstack-swift | 08:28 | |
| *** I has joined #openstack-swift | 08:40 | |
| *** I is now known as Guest10461 | 08:40 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Update EC Support on how to build an EC ring with replicas count https://review.openstack.org/223304 | 08:41 |
| *** haigang has joined #openstack-swift | 08:47 | |
| Guest10461 | #list | 08:55 |
| *** Guest10461 has quit IRC | 08:56 | |
| *** haigang has quit IRC | 08:56 | |
| *** haomaiwa_ has quit IRC | 08:59 | |
| *** haomaiwang has joined #openstack-swift | 09:05 | |
| openstackgerrit | Mahati Chamarthy proposed openstack/python-swiftclient: Add headers parameter https://review.openstack.org/223489 | 09:06 |
| *** mahatic has joined #openstack-swift | 09:07 | |
| openstackgerrit | Merged openstack/swift: Fix typo in Deployment Guide and add some formatting https://review.openstack.org/223356 | 09:08 |
| *** aix has quit IRC | 09:09 | |
| *** haomaiwang has quit IRC | 09:23 | |
| *** haomaiwa_ has joined #openstack-swift | 09:33 | |
| *** kota_ has quit IRC | 09:49 | |
| *** bill_az has joined #openstack-swift | 09:53 | |
| *** aix has joined #openstack-swift | 09:57 | |
| *** haomaiwa_ has quit IRC | 10:01 | |
| *** haomaiwang has joined #openstack-swift | 10:01 | |
| *** haomaiwang has quit IRC | 10:07 | |
| *** DericHorn-HP has joined #openstack-swift | 10:09 | |
| *** chlong has quit IRC | 10:09 | |
| *** haomaiwa_ has joined #openstack-swift | 10:10 | |
| *** chlong has joined #openstack-swift | 10:11 | |
| *** m_kazuhiro has quit IRC | 10:13 | |
| *** hezhiqiang has quit IRC | 10:14 | |
| *** baojg has quit IRC | 10:16 | |
| *** baojg has joined #openstack-swift | 10:16 | |
| *** baojg has quit IRC | 10:21 | |
| *** haomaiw__ has joined #openstack-swift | 10:22 | |
| *** haomaiwa_ has quit IRC | 10:24 | |
| *** mahatic has quit IRC | 10:25 | |
| *** SkyRocknRoll has quit IRC | 10:29 | |
| *** flwang has quit IRC | 10:31 | |
| *** asettle has quit IRC | 10:31 | |
| *** mahatic has joined #openstack-swift | 10:31 | |
| *** flwang has joined #openstack-swift | 10:32 | |
| *** flwang1 has joined #openstack-swift | 10:37 | |
| flwang1 | cschwede: ping | 10:37 |
| *** ujjain- has quit IRC | 10:52 | |
| *** ujjain has joined #openstack-swift | 10:53 | |
| *** ujjain has joined #openstack-swift | 10:53 | |
| *** DericHorn-HP has quit IRC | 10:55 | |
| *** haomaiw__ has quit IRC | 11:01 | |
| *** haomaiwang has joined #openstack-swift | 11:01 | |
| *** trifon_ has quit IRC | 11:08 | |
| *** trifon has joined #openstack-swift | 11:13 | |
| *** ccavanna_ has quit IRC | 11:19 | |
| *** mahatic has quit IRC | 11:20 | |
| *** mahatic has joined #openstack-swift | 11:24 | |
| *** mahatic has quit IRC | 11:27 | |
| *** mahatic has joined #openstack-swift | 11:28 | |
| *** hezhiqiang has joined #openstack-swift | 11:37 | |
| *** kei_yama has quit IRC | 11:40 | |
| *** haomaiwang has quit IRC | 11:42 | |
| *** cdelatte has quit IRC | 11:46 | |
| *** mahatic has quit IRC | 11:51 | |
| *** mahatic has joined #openstack-swift | 11:52 | |
| *** nakagawamsa has quit IRC | 12:05 | |
| *** mahatic has quit IRC | 12:07 | |
| *** mahatic has joined #openstack-swift | 12:08 | |
| notmyname | good morning | 12:10 |
| * notmyname just finished reading emails and IRC buffers | 12:11 | |
| *** cdelatte has joined #openstack-swift | 12:13 | |
| *** delattec has joined #openstack-swift | 12:14 | |
| *** delattec has quit IRC | 12:15 | |
| *** cdelatte has quit IRC | 12:18 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:20 | |
| *** km has quit IRC | 12:24 | |
| *** annegentle has joined #openstack-swift | 12:25 | |
| *** cdelatte has joined #openstack-swift | 12:25 | |
| * notmyname is off for another day of talking about swift | 12:26 | |
| *** MVenesio has joined #openstack-swift | 12:32 | |
| *** dmorita has quit IRC | 12:35 | |
| *** changbl has quit IRC | 12:36 | |
| *** mahatic has quit IRC | 12:38 | |
| *** DericHorn-HP has joined #openstack-swift | 12:41 | |
| mattoliverau | notmyname: morning and have fun. | 12:59 |
| * mattoliverau is off to bed.. and yes my phone is on silent :P | 12:59 | |
| *** tongli has joined #openstack-swift | 13:02 | |
| *** pberis1 has joined #openstack-swift | 13:03 | |
| *** dustins has joined #openstack-swift | 13:07 | |
| *** resker has joined #openstack-swift | 13:10 | |
| *** DericHorn-HP has quit IRC | 13:11 | |
| *** hrou has joined #openstack-swift | 13:12 | |
| *** esker has quit IRC | 13:14 | |
| *** dustins has quit IRC | 13:18 | |
| *** SkyRocknRoll has quit IRC | 13:20 | |
| *** dustins has joined #openstack-swift | 13:23 | |
| *** janonymous has joined #openstack-swift | 13:27 | |
| *** wbhuber has joined #openstack-swift | 13:29 | |
| *** ppai has quit IRC | 13:38 | |
| *** jkugel has joined #openstack-swift | 13:39 | |
| *** annegentle has quit IRC | 13:46 | |
| *** resker has quit IRC | 13:47 | |
| *** esker has joined #openstack-swift | 13:47 | |
| *** mahatic has joined #openstack-swift | 13:48 | |
| *** dustins has quit IRC | 13:59 | |
| *** haypo has joined #openstack-swift | 14:00 | |
| *** dustins has joined #openstack-swift | 14:03 | |
| hrou | Hey acoles, around ? | 14:07 |
| *** annegentle has joined #openstack-swift | 14:08 | |
| *** jrichli has joined #openstack-swift | 14:15 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 14:19 |
| *** david-lyle has quit IRC | 14:20 | |
| acoles | hrou: hi | 14:21 |
| hrou | Hey acoles ! I think I have an answer to my question, its fairly generic but: Regarding if-(not)-match (i.e. conditional get) functionality, do you always need to supply an object (i.e. you can't do if-match /<accnt>/<cont>/ with a etag), I poked around in the code and it seems we'll only compare when we have an object in hand at the object server. | 14:22 |
| *** silor has joined #openstack-swift | 14:23 | |
| hrou | You can always supply the header, but when you do it and query a container for example, we seem to return everything regardless. | 14:23 |
| hrou | * swift client will support it but internally its a head + get with the etag on all the objects individually | 14:24 |
| acoles | hrou: yes just objects, i think the code path is in swift.common.swob.Response - the constructor takes a conditional_response flag that looks to only ever be True in the object path | 14:29 |
| hrou | acoles, that's exactly what I was looking at : ) | 14:30 |
| hrou | acoles, Ok that's somewhat good news, though I wonder if that may ever change in the future. | 14:30 |
| acoles | hrou: you got it then :) | 14:30 |
| jrichli | acoles: hrou and I want to make sure that there is always client provided info on all if-(not)-match type requests that would tell the keymaster exactly which key was needed in order to possibly provide the correct comparison. | 14:31 |
| hrou | acoles, by change I mean, we simply send the request to the container for all objects and it passes down the etag to the underlying object server(s), but I doubt such functionality exists (it'd most likely be a head request first) which is Ok as long as the keymaster comes after this (so we can get the correct key). | 14:32 |
| hrou | jrichli, yep exactly | 14:32 |
| jrichli | acoles: for example, if there was a way to ask "return to me any object that has this etag", then there would be no way to know which key to use to encrypt the client provided etag to possibly have a successful comparison. | 14:34 |
| *** annegentle has quit IRC | 14:36 | |
| *** annegentle has joined #openstack-swift | 14:37 | |
| acoles | jrichli: hrou is that funtionality did exist, how would it work today? it would need a container list to find matching object, then do a GET for the object. That would work with crypto I think ? contaner key is used to decrypt etags in container listing, then object path would be know so object key can be found. | 14:38 |
| * acoles may be missing your point | 14:38 | |
| acoles | s/is/if/ at the start ^^ | 14:39 |
| *** chsc has joined #openstack-swift | 14:39 | |
| hrou | acoles, thanks acoles, no that was my thought as well ! i.e. if that functionality were to be implemented it would be done so as you describe, which is fine for crypto | 14:40 |
| hrou | acoles, jrichli, I think the one concern that still exists is something I highlighted in the etherpad, but its a rather convulted | 14:40 |
| hrou | acoles, jrichli something like: say you put object 1 on path X (client saves etag 1), another client puts object 2 on the same path X ... the keymaster may choose to use a different KEY for this etag (if it picks keys per path or per container that won't be the case). | 14:41 |
| acoles | hrou: if the object key is random rather than a function of object path then you would need to fetch it from object metadata first before doing the conditional GET. | 14:43 |
| acoles | hrou: and yes, that would be prone to inconsistency between replicas | 14:43 |
| acoles | hrou: even if you HEAD the object for its key, you may then send GET to a different replica which has a different key | 14:44 |
| hrou | acoles, yep exactly, which in theory is applicable to scenarios other than conditional gets. | 14:44 |
| jrichli | acoles: given that container listing is acceptable in the handling on the way in, maybe another option to our if-not-match issues is to put the crypto-meta in the container listing - encrypted with container key | 14:45 |
| jrichli | that way, we could still remain flexible on the crypto-alg used for etag | 14:46 |
| *** jlhinson has joined #openstack-swift | 14:46 | |
| acoles | jrichli: are we still talking about a hypothetical future feature? | 14:47 |
| hrou | jrichli, what do you mean by acceptable though ? Currently there is no head request for a "if .. match", the "head" was a potential 'future' implementation to support my hypothetical scenario : ) | 14:47 |
| jrichli | acoles: do you mean the part about being flexible on crypto-alg? | 14:48 |
| acoles | jrichli: i don't think i mean that :) this started with a hypothetical conditional GET for any object in a container | 14:49 |
| acoles | yeah, what hrou said | 14:49 |
| jrichli | oh, right. And I liked your answer - I wouldn't have thought of it because I typically think we should not be doing requests from the middleware - but i remember now, container requests are better than obj reqs | 14:50 |
| acoles | doing a container listing to get the object etag or (hypothetically) the object key has same inconsistency issue as doing an object HEAD - the container listing may not be consistent with the state of the object replica you then send a GET to | 14:50 |
| acoles | container info lookups are "better" because they often do not incur a request when info is cached. But they are not better in terms of any guarantee of consistency with object state. | 14:51 |
| hrou | acoles, I'd probably state that's a generic concern that exists even today with the current implementation (not related to conditional get) e.g. re-put an object, try to get, if keymaster creates a new key it better manage which one to send due to the eventual consistency nature of the system | 14:52 |
| acoles | hrou: trivial keymaster always creates same key for same path | 14:53 |
| acoles | and its only an issue when you need the key before doing the GET - normal GETs would return the key if it were stored with the object | 14:53 |
| hrou | acoles, yep ! Sorry I meant for another implementation that doesn't base it on paths (I think the IBM keymaster was going down the route, jrichli ?) But I think we discussed some solutions for that during the hackathon | 14:54 |
| hrou | acoles, yep that's true ! Though I'm not sure where the majority of keymasters will be storing keys, may not reside with the object (or swift in general) | 14:55 |
| jrichli | hrou: yes. you can see Trello cards in column 'To Do' for Rare Cases to see if all the ideas you are talking about are represented there | 14:56 |
| hrou | jrichli, thanks yea ! I recall the card for it. | 14:56 |
| hrou | Ok, I feel better about conditional gets as a whole, thanks jrichli, acoles ! | 14:56 |
| jrichli | acoles hrou: this is getting into "real" keymaster talk. I was thinking of setting up a new etherpad for thoughts on what a real keymaster might look like, and what issues it may have | 14:57 |
| hrou | jrichli, yea that's a really good idea. | 14:57 |
| jrichli | or, i could tack it onto the existing pad, but its getting sort of full | 14:57 |
| jrichli | actually, come to think of it, that was acoles idea | 14:58 |
| jrichli | when i talked with him about re-key | 14:58 |
| hrou | jrichli, though that kinda of bothers me : ) See we're trying to create a generic implemenation that can satisfy any and all mythical keymasters, but clearly they're implementation has a strong effect on the reset of crypto. | 14:58 |
| *** minwoob has joined #openstack-swift | 14:59 | |
| hrou | jrichli, yep I remember that @ the hackathon (I think that's when I had to leave : - ) | 14:59 |
| acoles | i don't remember that ! | 14:59 |
| jrichli | But yes, a lot of our thinking has been about a keymaster that uses swift itself to store the keys. since there aren't many cloud-scalable choices for storing a different key for each object, and key derivation was *questionable* to some people | 14:59 |
| hrou | acoles, haha ! | 14:59 |
| hrou | jrichli, right yea that makes sense, though we're proposing to do key derivation now ; ) But that's for the specific obj / meta / etag keys. | 15:00 |
| *** pberis1 has quit IRC | 15:01 | |
| hrou | jrichli, seperate etherpad could work but I think it may effect things other than the keymaster (i.e. require changes to other parts of crypto), no big deal either way for now! | 15:01 |
| jrichli | one of the people here that disliked the idea of key derivation, also doesn't really care about the etag being revealed. I explained that having the etags would mean that you have an easy way to verify whether your object body attach was successful | 15:01 |
| jrichli | s/attach/attack/ | 15:02 |
| hrou | jrichli, was the concern derived keys in the sense of "you have one key for a container" and derive the rest ? I can understand that, we're proposing something much more granular. | 15:02 |
| *** SkyRocknRoll has joined #openstack-swift | 15:03 | |
| hrou | jrichli, i.e. we're just doing it within swift to ensure we have separate keys for each crpyto unit. They may be Ok with that aspect ? | 15:03 |
| jrichli | again, this individual doesn't have issues with the current proposal, since its only about etag | 15:03 |
| hrou | jrichli, ah gotcha, yep, that makes sense. | 15:04 |
| hrou | jrichli, but yea we could start up some calls / conversations about the 'real' keymaster and the effect it may have on the rest of the solution ? | 15:04 |
| jrichli | hrou: sounds good. I think the new etherpad would be a good start. | 15:05 |
| hrou | jrichli, yep! If we find anything that effects the rest of crypto we can just put a link into the other one. | 15:05 |
| * jrichli is going to cafeteria for free starbucks beverage | 15:07 | |
| *** kevinc_ has joined #openstack-swift | 15:11 | |
| *** david-lyle has joined #openstack-swift | 15:13 | |
| *** then3rd has quit IRC | 15:18 | |
| *** SkyRocknRoll has quit IRC | 15:21 | |
| *** garthb has joined #openstack-swift | 15:33 | |
| *** robefran has joined #openstack-swift | 15:40 | |
| *** gyee has joined #openstack-swift | 15:53 | |
| *** rohit_ has joined #openstack-swift | 15:57 | |
| *** esker has quit IRC | 16:12 | |
| *** janonymous_ has joined #openstack-swift | 16:20 | |
| briancline | morning | 16:20 |
| briancline | might anyone have a few minutes to review patch 219617? | 16:20 |
| patchbot | briancline: https://review.openstack.org/#/c/219617/ | 16:20 |
| *** annegent_ has joined #openstack-swift | 16:22 | |
| *** jistr has quit IRC | 16:24 | |
| *** annegentle has quit IRC | 16:24 | |
| *** annegentle has joined #openstack-swift | 16:28 | |
| *** annegentle has quit IRC | 16:30 | |
| *** annegentle has joined #openstack-swift | 16:31 | |
| *** jordanP has quit IRC | 16:31 | |
| *** annegent_ has quit IRC | 16:31 | |
| *** breitz has joined #openstack-swift | 16:32 | |
| *** breitz has quit IRC | 16:34 | |
| *** breitz has joined #openstack-swift | 16:35 | |
| *** breitz has quit IRC | 16:37 | |
| *** breitz has joined #openstack-swift | 16:38 | |
| *** aix has quit IRC | 16:39 | |
| *** kutija has joined #openstack-swift | 16:40 | |
| *** rledisez has quit IRC | 16:41 | |
| *** janonymous has quit IRC | 16:43 | |
| *** breitz has quit IRC | 16:44 | |
| *** janonymous_ has quit IRC | 16:46 | |
| *** breitz has joined #openstack-swift | 16:48 | |
| *** cdelatte has quit IRC | 16:48 | |
| *** fifieldt_ has quit IRC | 16:49 | |
| *** geaaru has quit IRC | 17:04 | |
| *** cdelatte has joined #openstack-swift | 17:07 | |
| *** delattec has joined #openstack-swift | 17:07 | |
| *** changbl has joined #openstack-swift | 17:17 | |
| *** hezhiqiang has quit IRC | 17:19 | |
| *** oc714|2 has left #openstack-swift | 17:26 | |
| *** annegentle has quit IRC | 17:39 | |
| *** annegentle has joined #openstack-swift | 17:47 | |
| *** chenhuayi has quit IRC | 17:49 | |
| *** chenhuayi has joined #openstack-swift | 17:50 | |
| *** annegent_ has joined #openstack-swift | 17:56 | |
| *** annegentle has quit IRC | 18:00 | |
| *** annegent_ has quit IRC | 18:04 | |
| *** annegentle has joined #openstack-swift | 18:05 | |
| *** annegentle has quit IRC | 18:11 | |
| mahatic | jrichli: hrou the error from 'gate-swiftclient-dvsm-functional' on this patch 223489 happen when the encryption is on. But I'm not sure from where is the gate picking encryption. any ideas? | 18:11 |
| patchbot | mahatic: https://review.openstack.org/#/c/223489/ | 18:11 |
| *** bill_az has quit IRC | 18:13 | |
| *** annegentle has joined #openstack-swift | 18:13 | |
| *** cdelatte has quit IRC | 18:35 | |
| *** delattec has quit IRC | 18:35 | |
| clayg | hello | 18:38 |
| clayg | briancline: I'll take a look | 18:38 |
| jrichli | mahatic: I don't know of a reason that the gate would pick-up encryption for a change against the python-swiftclient repo. Are you sure that is happening? | 18:39 |
| *** haypo has left #openstack-swift | 18:40 | |
| mahatic | jrichli: I don't have an error on my local repo when I take off encryption from proxy conf. Otherwise it threw an error on etag mismatch, that got me confused | 18:42 |
| jrichli | mahatic: the error I am seeing from the gate is not about an etag mismatch. It is getting unauthorized on a few container listing tests. | 18:43 |
| mahatic | jrichli: right, test_list_container. that's where i have etag mismatch on my repo | 18:44 |
| jrichli | mahatic: are you trying to find out why the gate is failing (the authorization error), or why the changes locally have an etag issue when encryption is on? | 18:45 |
| mahatic | jrichli: the first (which lead me to the latter) | 18:46 |
| jrichli | mahatic: ok. focusing on the gate issue, I think it would be best to try to reproduce that locally - with encryption out of the pipeline. I don't think there is any way that gate can have encryption code to run. | 18:48 |
| mahatic | jrichli: :) I did that and the result is etag mismatch error on the same test | 18:48 |
| mahatic | i.e. test_list_container | 18:48 |
| *** dustins has quit IRC | 18:49 | |
| mahatic | jrichli: oh w/o encryption on pipeline, no errors. Could you cross check that? | 18:49 |
| jrichli | sure, i will give it a spin. | 18:49 |
| mahatic | great, thanks! | 18:49 |
| *** delattec has joined #openstack-swift | 18:50 | |
| *** cdelatte has joined #openstack-swift | 18:50 | |
| *** wshao has joined #openstack-swift | 18:50 | |
| *** hezhiqiang has joined #openstack-swift | 18:50 | |
| clayg | so gross that head_object in swiftclient doesn't have the headers param (how's a brother supposed to if-match!?) - re: patch 223489 | 18:54 |
| patchbot | clayg: https://review.openstack.org/#/c/223489/ | 18:54 |
| clayg | torgomatic: timburke: you guys should also put a vote on patch 223489 | 18:54 |
| patchbot | clayg: https://review.openstack.org/#/c/223489/ | 18:54 |
| mahatic | clayg: thanks for looking at that ! | 18:55 |
| clayg | mahatic: didn't take long :\ | 18:56 |
| clayg | mahatic: let me know when you push something else! | 18:56 |
| clayg | mahatic: thanks for working on it! | 18:56 |
| *** hezhiqiang has quit IRC | 18:56 | |
| clayg | what's hamdi's handle again? | 18:56 |
| mahatic | clayg: sure thing. and yes this needs tests! | 18:57 |
| jrichli | clayg: hrou | 18:57 |
| mahatic | clayg: hrou there | 18:57 |
| clayg | hrou: kudos on you too - thanks for working to make swiftclient better! | 18:57 |
| *** dustins has joined #openstack-swift | 18:59 | |
| acoles | mahatic: see my review on patch 223489, tox -e func works for me when i pass headers=headers | 19:00 |
| patchbot | acoles: https://review.openstack.org/#/c/223489/ | 19:00 |
| clayg | acoles: how can I help review optomistic gets? diskfile dict change? | 19:02 |
| mahatic | acoles: just saw that, thanks much! it makes sense. Will pull it down and verify | 19:03 |
| *** esker has joined #openstack-swift | 19:04 | |
| *** esker has joined #openstack-swift | 19:05 | |
| acoles | clayg: yes that would be a good place to start. thanks! patch 222706 | 19:05 |
| patchbot | acoles: https://review.openstack.org/#/c/222706/ | 19:05 |
| *** gyee has quit IRC | 19:05 | |
| *** esker has quit IRC | 19:05 | |
| *** hezhiqiang has joined #openstack-swift | 19:06 | |
| *** esker has joined #openstack-swift | 19:06 | |
| acoles | clayg: then patch 215276 i marked as WIP but what is there might be worth looking over - its WIP because I need to add the proxy side changes | 19:08 |
| patchbot | acoles: https://review.openstack.org/#/c/215276/ | 19:08 |
| acoles | clayg: been thinking I should change X-Durable to X-Backend-Durable | 19:09 |
| acoles | clayg: i should have the optimistic proxy side of things ready tomorrow. | 19:15 |
| *** hezhiqiang has quit IRC | 19:15 | |
| *** acoles is now known as acoles_ | 19:16 | |
| *** mahatic has quit IRC | 19:17 | |
| *** silor1 has joined #openstack-swift | 19:28 | |
| *** dustins has quit IRC | 19:30 | |
| *** dustins has joined #openstack-swift | 19:30 | |
| *** silor has quit IRC | 19:32 | |
| *** silor1 is now known as silor | 19:32 | |
| clayg | acoles_: oh yeah -Backend- FTW | 19:33 |
| clayg | i don't really know what to say to the nobarriers guy on the ML | 19:36 |
| clayg | 'ok so, maybe you suggest we change it to "nobariers is probably faster and maybe what you want on some hardware but YMMV" | 19:36 |
| *** DericHorn-HP has joined #openstack-swift | 19:38 | |
| *** hezhiqiang has joined #openstack-swift | 19:40 | |
| *** flwang1 has quit IRC | 19:48 | |
| *** silor1 has joined #openstack-swift | 19:49 | |
| *** annegentle has quit IRC | 19:50 | |
| *** devlaps has joined #openstack-swift | 19:51 | |
| *** silor has quit IRC | 19:52 | |
| *** silor1 is now known as silor | 19:52 | |
| *** hezhiqiang has quit IRC | 19:56 | |
| * clayg puts using rclone on my list | 19:59 | |
| *** hezhiqiang has joined #openstack-swift | 20:07 | |
| *** silor1 has joined #openstack-swift | 20:07 | |
| *** wshao has quit IRC | 20:08 | |
| *** gyee has joined #openstack-swift | 20:08 | |
| *** silor has quit IRC | 20:09 | |
| *** silor1 is now known as silor | 20:09 | |
| *** proteusguy_ has quit IRC | 20:12 | |
| *** tsimp06 has joined #openstack-swift | 20:14 | |
| *** robefran has quit IRC | 20:14 | |
| *** lcurtis has joined #openstack-swift | 20:20 | |
| *** hezhiqiang has quit IRC | 20:20 | |
| *** dustins_ has joined #openstack-swift | 20:31 | |
| briancline | TIL what rclone is. very cool | 20:33 |
| *** dustins has quit IRC | 20:34 | |
| *** silor has quit IRC | 20:35 | |
| *** proteusguy_ has joined #openstack-swift | 20:36 | |
| *** annegentle has joined #openstack-swift | 20:37 | |
| openstackgerrit | Richard Hawkins proposed openstack/swift: Fix object name length check. https://review.openstack.org/223797 | 20:38 |
| *** dustins_ has quit IRC | 20:40 | |
| hurricanerix | notmyname: Should I have submitted a bug before submitting a patch? ^^^ | 20:43 |
| clayg | hurricanerix: most of the time, yes - it complete's the circle of life | 20:45 |
| clayg | is that constraint only enforced on PUT/COPY - that sort of thing - or do we lock data in clusters!? | 20:46 |
| clayg | hurricanerix: ^ | 20:47 |
| hurricanerix | clayg: That is a good question, I was only concerned with PUT/COPY when writing the patch. | 20:47 |
| hurricanerix | clayg: let me check a GET | 20:48 |
| clayg | briancline: rclone is cool because it's in golang - golang makes everything cooler | 20:48 |
| openstackgerrit | Thiago da Silva proposed openstack/swift: WIP: new attempt at single-process https://review.openstack.org/159285 | 20:48 |
| clayg | hurricanerix: I would have swore there was an existing functest for object name length - crazy | 20:50 |
| hurricanerix | clayg: do you remember if it checked for the name containing percent encoded values? | 20:51 |
| clayg | hurricanerix: no i'm sure it wouldn't have or your change would have been needed :) | 20:51 |
| hurricanerix | because if it is just 'aaaa....' then it worked as is. | 20:51 |
| clayg | but it coudl probably be updated w/o having to add that second new one? | 20:51 |
| *** cdelatte has quit IRC | 20:51 | |
| *** delattec has quit IRC | 20:51 | |
| hurricanerix | clayg: yeah, I'll look around and if I can find it, I'll clean my tests up. | 20:52 |
| clayg | tests.TestFile.testNameLimit | 20:52 |
| clayg | hurricanerix: I just grepped for max_object_name_length | 20:53 |
| hurricanerix | lol, whoops =) | 20:53 |
| *** changbl has quit IRC | 20:55 | |
| hurricanerix | clayg: looks like you can still GET | 20:55 |
| clayg | hurricanerix: that's good right!? | 20:55 |
| clayg | hurricanerix: probably wouldn't hurt to have a unitest for it tho | 20:56 |
| clayg | well... it might hurt a little | 20:56 |
| hurricanerix | clayg: I think so, I don't think I would want the fix to affect people who might have already accidentally exceeded the limit. | 20:56 |
| *** pgbridge has quit IRC | 20:56 | |
| clayg | it wouldn't hurt *me* tho - cause I woudln't be the one writing it! | 20:56 |
| clayg | hurricanerix: +1 | 20:56 |
| hurricanerix | But they might get a surprise if they try to change the object. | 20:57 |
| clayg | *suprise*! | 20:57 |
| *** wshao has joined #openstack-swift | 20:57 | |
| hurricanerix | clayg: or if they try to update metadata. | 20:58 |
| *** hrou has quit IRC | 20:59 | |
| clayg | but I agree in principle that constraints are there to protect the operator - if you can blow up the lenght limit by 3x by making an object with lots of ' ' in the name that's not really honoring the deployers constraints - so I think it's a reasonable enough change - deployer could up the limit if they surprise anyone? | 20:59 |
| hurricanerix | true | 20:59 |
| clayg | then it'd be more like setting the limit in the conf file to the *real* limit | 20:59 |
| *** hrou has joined #openstack-swift | 20:59 | |
| *** mragupat has joined #openstack-swift | 20:59 | |
| *** flwang1 has joined #openstack-swift | 21:02 | |
| *** garthb has quit IRC | 21:03 | |
| clayg | notmyname: I'm really not loving the new gerrit view :\ | 21:06 |
| *** DericHorn-HP has quit IRC | 21:08 | |
| *** DericHorn-HP has joined #openstack-swift | 21:17 | |
| *** MVenesio has quit IRC | 21:20 | |
| *** hezhiqiang has joined #openstack-swift | 21:21 | |
| *** mragupat has quit IRC | 21:22 | |
| *** mragupat has joined #openstack-swift | 21:22 | |
| *** hezhiqiang has quit IRC | 21:25 | |
| briancline | the name length fix is an interesting one | 21:28 |
| briancline | how do unicode characters get counted? | 21:29 |
| *** rohit_ has quit IRC | 21:30 | |
| *** wbhuber has quit IRC | 21:30 | |
| *** garthb has joined #openstack-swift | 21:33 | |
| *** flwang1 has left #openstack-swift | 21:36 | |
| *** kevinc_ has quit IRC | 21:36 | |
| hurricanerix | briancline: I need to verify, but I would guess like this: In [2]: urllib.quote('☃') | 21:40 |
| hurricanerix | Out[2]: '%E2%98%83', so counted as 9 chars. | 21:40 |
| clayg | f'ing unicode | 21:42 |
| torgomatic | I'd hope for 3, because that's the length of the UTF-8 representation | 21:43 |
| torgomatic | but, bigger question: what's the point of the name check? what would we lose if we took it out? | 21:43 |
| clayg | but not all utf-8 bytes are url-safe | 21:43 |
| torgomatic | (i'm not proposing to take it out) | 21:43 |
| clayg | torgomatic: people would put zero byte object there were like 'name data' - and then do prefix listings | 21:44 |
| clayg | ... or something | 21:44 |
| torgomatic | clayg: oh, so it's a "hey jerks, knock it off" kind of thing | 21:44 |
| clayg | yeah I think so - or maybe just in general to keep the containers rows less than a few K wide | 21:45 |
| *** pberis has joined #openstack-swift | 21:45 | |
| clayg | there's probably no limit to how big your content-length can be | 21:45 |
| *** jrichli has quit IRC | 21:45 | |
| clayg | jerks | 21:45 |
| torgomatic | well, there's the header-line length limit | 21:46 |
| torgomatic | that'll get everything but the name | 21:46 |
| clayg | stupid name | 21:46 |
| clayg | ascii had more than enough printable characters for anyone | 21:47 |
| clayg | and you know how wide they were!? what char! | 21:49 |
| clayg | *one | 21:49 |
| *** tongli has quit IRC | 21:51 | |
| clayg | q | 22:02 |
| *** hrou has quit IRC | 22:03 | |
| *** _hrou_ has joined #openstack-swift | 22:03 | |
| _hrou_ | thanks clayg ! Just getting caught up, and great work acoles_, all for find that gate issue with the patch. | 22:04 |
| _hrou_ | * s/find/finding | 22:05 |
| *** jkugel has quit IRC | 22:06 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: fixups for acoles https://review.openstack.org/223854 | 22:11 |
| *** asettle has joined #openstack-swift | 22:24 | |
| *** pberis has quit IRC | 22:27 | |
| openstackgerrit | Alexandra Settle proposed openstack/swift: Moving DLO functionality doc to middleware https://review.openstack.org/219991 | 22:27 |
| mattoliverau | Morning | 22:30 |
| asettle | Morning o/ | 22:30 |
| *** lcurtis has quit IRC | 22:32 | |
| *** ChanServ sets mode: +v mattoliverau | 22:42 | |
| *** asettle has quit IRC | 22:43 | |
| *** DericHorn-HP has quit IRC | 22:51 | |
| *** darrenc is now known as darrenc_afk | 22:52 | |
| *** aix has joined #openstack-swift | 22:52 | |
| *** mragupat has quit IRC | 22:56 | |
| *** zhill has joined #openstack-swift | 22:57 | |
| *** wshao has quit IRC | 23:01 | |
| *** annegentle has quit IRC | 23:03 | |
| *** km has joined #openstack-swift | 23:04 | |
| *** darrenc_afk is now known as darrenc | 23:06 | |
| *** openstackgerrit has quit IRC | 23:16 | |
| *** openstackgerrit has joined #openstack-swift | 23:16 | |
| *** david-lyle has quit IRC | 23:17 | |
| *** chsc has quit IRC | 23:17 | |
| *** kei_yama has joined #openstack-swift | 23:27 | |
| *** _hrou_ has quit IRC | 23:33 | |
| *** breitz has quit IRC | 23:41 | |
| mattoliverau | And the channel goes quiet when I arrive. Do I need to change my nick to be about to stalk you all?.. again :P | 23:42 |
| *** ho has joined #openstack-swift | 23:47 | |
| tdasilva | mattoliverau: hey!! you just need to come back to our timezone :-) | 23:51 |
| *** jlhinson has quit IRC | 23:51 | |
| mattoliverau | tdasilva: lol, true.. or you all come to mine, say next month? | 23:52 |
| tdasilva | mattoliverau: sounds about right :-) my agenda is quite busy for the beginning of the month, but I should be there at the end of the month ;) | 23:53 |
| *** aix has quit IRC | 23:53 | |
| mattoliverau | tdasilva: great, lets make a week of it, we'll talk swift, maybe go grab some sushi ;P | 23:55 |
| tdasilva | mattoliverau: deal! | 23:55 |
| tdasilva | mattoliverau: feeling better? | 23:57 |
| mattoliverau | tdasilva: yeah much, still not 100% but enough to work faster. Managed to fumble round and get the first version of the container-sharder daemon completed (inline with last hackathon). So hasn't been a wasted week :) | 23:59 |
| *** wshao has joined #openstack-swift | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!