| *** hoonetorg has quit IRC | 00:35 | |
| *** hoonetorg has joined #openstack-swift | 00:37 | |
| *** hoonetorg has quit IRC | 00:40 | |
| *** hoonetorg has joined #openstack-swift | 00:40 | |
| *** two_tired has joined #openstack-swift | 01:48 | |
| kota_ | good morning | 03:48 |
|---|---|---|
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: py3: adapt the account server completely https://review.openstack.org/613505 | 04:39 |
| *** two_tired has quit IRC | 04:47 | |
| *** ccamacho has quit IRC | 06:24 | |
| *** spsurya has joined #openstack-swift | 06:39 | |
| *** mahatic has quit IRC | 06:51 | |
| *** mahatic has joined #openstack-swift | 06:52 | |
| *** ChanServ sets mode: +v mahatic | 06:52 | |
| *** gkadam has joined #openstack-swift | 07:02 | |
| *** mahatic has quit IRC | 07:12 | |
| *** pcaruana has joined #openstack-swift | 07:20 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: py3: port encryption https://review.openstack.org/601454 | 07:30 |
| *** pcaruana has quit IRC | 07:34 | |
| *** psachin has joined #openstack-swift | 07:37 | |
| *** pcaruana has joined #openstack-swift | 07:40 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: py3: adapt the account server completely https://review.openstack.org/613505 | 07:53 |
| *** ccamacho has joined #openstack-swift | 07:54 | |
| *** zaitcev has quit IRC | 08:02 | |
| *** e0ne has joined #openstack-swift | 09:24 | |
| *** mahatic has joined #openstack-swift | 09:41 | |
| *** ChanServ sets mode: +v mahatic | 09:41 | |
| *** gkadam has quit IRC | 11:56 | |
| *** gkadam has joined #openstack-swift | 12:14 | |
| *** psachin has quit IRC | 12:41 | |
| *** zaitcev has joined #openstack-swift | 14:25 | |
| *** ChanServ sets mode: +v zaitcev | 14:25 | |
| rledisez | hi | 15:53 |
| rledisez | simple question: right now, the ring is limited to 2 ** 16 devices (65536). how hard would it be to increase this limit? (eg: 2 ** 32) | 15:55 |
| rledisez | it seems quite easy quickly reading the ring (serialize_v1 method and NONE_DEV), but I might be missing something | 15:55 |
| *** itlinux has quit IRC | 15:57 | |
| *** ccamacho has quit IRC | 16:20 | |
| *** e0ne has quit IRC | 16:21 | |
| DHE | I thought it was available to around 32 or so, but it's user configurable at ring creation time | 16:30 |
| DHE | or am I confusing partitions and devices | 16:30 |
| rledisez | DHE: are you talking about the part power? yes, it's a number of bits that define the number of partitions in a cluster | 16:42 |
| *** itlinux has joined #openstack-swift | 16:43 | |
| rledisez | I'm talking about https://github.com/openstack/swift/blob/master/swift/common/ring/builder.py#L43 and https://github.com/openstack/swift/blob/master/swift/common/ring/ring.py#L148 | 16:43 |
| DHE | I see... | 16:44 |
| DHE | so device number 65535 can't exist because it would be confused for a gap in the ring profile... | 16:44 |
| rledisez | well, forget about !I, it's not related, but NONE_DEV, yes, it's a limit | 16:45 |
| rledisez | DHE: exactly | 16:46 |
| rledisez | we are not there yet, but we are starting to think of it. better be prepared ;) | 16:48 |
| *** gkadam has quit IRC | 16:50 | |
| DHE | I'm mildly jealous.... | 17:08 |
| tdasilva | heh | 17:26 |
| *** e0ne has joined #openstack-swift | 17:31 | |
| *** e0ne has quit IRC | 17:44 | |
| *** e0ne has joined #openstack-swift | 17:57 | |
| timburke | rledisez: i *think* it'd be as "simple" as adding a key to the JSON dict at the front, like device_bits or something, then over in deserialize_v1 we'd validate that it's either 16 or 32 and use that to switch between H and I when we load up the array.array | 18:03 |
| timburke | we could get pretty far into that space before we'd need to start worrying about the 32-bit limit on the device dict, too -- but even then we could just shift the magic to something like R2NG and start using a 64-bit int for the json size | 18:05 |
| timburke | biggest downside is that we'd be doubling our memory footprint, so maybe only switch to 32-bit if we *have to*? of course, as long as your part power's something sane, it still shouldn't be that bad (i think)... | 18:06 |
| DHE | wouldn't you need a part_power around 21 or so for such a cluster? | 18:14 |
| timburke | DHE: yeah -- and that's not so bad. i'm more worried about something inching up towards 30 -- as we are now, you're over a gig just to load up one ring with a pp of 29 | 18:19 |
| timburke | exponents are rough | 18:19 |
| zaitcev | timburke: did you see my changes to the "encryption" py3 patch? | 18:24 |
| timburke | yeah -- hoping to look through it this morning | 18:24 |
| timburke | should i +A if i'm happy? | 18:25 |
| zaitcev | yes | 18:25 |
| zaitcev | IMHO | 18:25 |
| zaitcev | Also, I found that my "finish the account" py3 thing didn't take advantage of json.loads monkey-patching for some reason, so I removed some code from there too. But that's less important. | 18:26 |
| timburke | funny -- i coulda sworn i'd pulled that out on my second push -- w/e | 18:33 |
| *** pcaruana has quit IRC | 18:47 | |
| *** e0ne has quit IRC | 19:22 | |
| *** itlinux has quit IRC | 20:39 | |
| *** AJaeger has joined #openstack-swift | 20:53 | |
| AJaeger | swift team, could you import your rocky translations, please? https://review.openstack.org/615081 | 20:54 |
| patchbot | patch 615081 - swift (stable/rocky) - Imported Translations from Zanata - 1 patch set | 20:54 |
| timburke | notmyname: ^^^ | 21:10 |
| *** itlinux has joined #openstack-swift | 21:13 | |
| timburke | i... what? what is this testing?? https://github.com/openstack/swift/blob/2.19.0/test/unit/common/test_utils.py#L1284-L1289 | 21:38 |
| *** rcernin has joined #openstack-swift | 21:38 | |
| tdasilva | mmm... | 21:40 |
| tdasilva | timburke: more importantly (IMO) i was wondering how did it get past review, turns out it probably didn't as it was in the initial commit... | 21:41 |
| tdasilva | was probably temp code never got finished... | 21:42 |
| timburke | yup... | 21:42 |
| timburke | 21:42 | |
| zaitcev | sio is clearly unused, the rest makes sure that NullLogger has a method that doesn't crash. | 22:07 |
| mattoliverau | morning | 22:18 |
| mattoliverau | @AJaeger because that's on the stable branch only notmyname has +2/+A writes there. on the swift team anyway. | 22:20 |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: encryption follow-up https://review.openstack.org/619109 | 22:27 |
| timburke | zaitcev: ^^^ | 22:31 |
| *** itlinux has quit IRC | 22:44 | |
| openstackgerrit | Merged openstack/swift master: py3: port encryption https://review.openstack.org/601454 | 22:45 |
| *** zaitcev has quit IRC | 23:56 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!