*** garthb has quit IRC | 00:02 | |
*** nadeem has quit IRC | 00:03 | |
*** diogogmt has quit IRC | 00:03 | |
*** Kennan has quit IRC | 00:16 | |
*** CR7 has quit IRC | 00:20 | |
*** chlong has joined #openstack-swift | 00:21 | |
*** lyrrad has quit IRC | 00:23 | |
*** bill_az has quit IRC | 00:25 | |
*** wbhuber has quit IRC | 00:29 | |
*** km has quit IRC | 00:29 | |
*** wbhuber has joined #openstack-swift | 00:32 | |
*** CR7 has joined #openstack-swift | 00:36 | |
*** CR7 has quit IRC | 00:40 | |
kota_ | morning | 00:40 |
---|---|---|
*** m_kazuhiro has joined #openstack-swift | 00:48 | |
*** wbhuber has quit IRC | 00:54 | |
*** sanchitmalhotra has joined #openstack-swift | 00:55 | |
*** sanchitmalhotra1 has quit IRC | 00:56 | |
*** gyee has quit IRC | 00:59 | |
*** ccavanna has quit IRC | 01:00 | |
*** ccavanna has joined #openstack-swift | 01:05 | |
*** sanchitmalhotra1 has joined #openstack-swift | 01:11 | |
*** sanchitmalhotra has quit IRC | 01:13 | |
*** amit213 has quit IRC | 01:24 | |
*** amit213 has joined #openstack-swift | 01:24 | |
*** hrou has joined #openstack-swift | 01:56 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 01:59 |
*** jrichli has joined #openstack-swift | 02:01 | |
*** wbhuber has joined #openstack-swift | 02:19 | |
clayg | where's peluse been? | 02:21 |
*** chlong has quit IRC | 02:27 | |
mattoliverau | kota_: morning | 02:28 |
mattoliverau | acoles_: retested patch, working well, just +A'ed. | 02:29 |
*** bill_az_ has quit IRC | 02:40 | |
*** chlong has joined #openstack-swift | 02:41 | |
*** wbhuber_ has joined #openstack-swift | 03:02 | |
*** wbhuber has quit IRC | 03:05 | |
*** wbhuber_ has quit IRC | 03:07 | |
*** wbhuber has joined #openstack-swift | 03:08 | |
*** wer has quit IRC | 03:09 | |
*** wer has joined #openstack-swift | 03:10 | |
openstackgerrit | Merged openstack/swift: Fix copy requests to service accounts in Keystone https://review.openstack.org/227133 | 03:10 |
*** wbhuber has quit IRC | 03:12 | |
*** trifon has joined #openstack-swift | 03:25 | |
*** jrichli has quit IRC | 03:25 | |
notmyname | hello, world | 03:26 |
*** links has joined #openstack-swift | 03:31 | |
*** trifon has quit IRC | 03:44 | |
*** kei_yama has quit IRC | 04:02 | |
*** kei_yama has joined #openstack-swift | 04:03 | |
*** klrmn1 has quit IRC | 04:22 | |
*** kei_yama has quit IRC | 04:35 | |
*** kei_yama has joined #openstack-swift | 04:36 | |
clayg | ok, one more run of the probetests I think | 04:42 |
clayg | acoles_: good news! my diff on patch 138498 is < 1000 lines! | 04:43 |
patchbot | clayg: https://review.openstack.org/#/c/138498/ - Add POST capability to ssync for .meta files | 04:43 |
*** mahatic has joined #openstack-swift | 04:43 | |
notmyname | clayg: I'm just about finished going through it, I think | 04:44 |
notmyname | well, "finished" is generous | 04:44 |
clayg | notmyname: I started with tcpdump and worked my way backwards from probetests to implementation to unittests ;) | 04:46 |
clayg | notmyname: acoles_ is good - it's a big change and there's really not that much slop | 04:46 |
notmyname | that is more thorough than I've done ;-) | 04:47 |
clayg | fuck | 04:47 |
notmyname | find something else? | 04:47 |
clayg | thinking of slop I realized I factored out some methods to functions and didn't add explict unittests for them :\ | 04:47 |
clayg | nah - just that constant nagging feeling I'm not at good as software development as most of the people around me | 04:47 |
notmyname | I very strongly sympathize with that | 04:48 |
clayg | *as | 04:48 |
clayg | well, let me slap a couple tests on there | 04:48 |
mahatic | hrou: hello! thanks for the review, if I understood it correcctly, you're already working on correcting the encoding way on this 194191 per timburke's comments and you want to see if that's gonna resolve? | 04:49 |
*** trifon has joined #openstack-swift | 04:51 | |
hrou | mahatic, yep ! I think I'll need to do more to resolve it though, timburke made some good points in a comment as well, so I'm going to look into those (but yea in other words don't worry about that failing test, and keep it using utf8 | 04:51 |
mahatic | hrou: I'm happy to continue looking into that once you're done with 194191 | 04:52 |
hrou | mahatic, awesome great ! Yea I should have something for that patch tomr | 04:53 |
mahatic | hrou: cool, I'm looking forward to that change | 04:55 |
mahatic | hrou: and we could really use acoles_ opinion here like you said | 04:56 |
clayg | and I found a bug in my fucking code too | 04:59 |
hrou | mahatic, yep, well, mind you that was before tim replied : ) | 04:59 |
*** m_kazuhiro has quit IRC | 04:59 | |
mahatic | hrou: yes, I missed that one actually | 05:00 |
clayg | weird that probetests didn't catch it - better run 'em again | 05:00 |
mahatic | timburke: thanks for the review! | 05:01 |
*** mragupat has joined #openstack-swift | 05:06 | |
mahatic | clayg: "most people" (re s/w dev) is a bit of a stretch ;) | 05:08 |
notmyname | clayg: I'm ready to look at your patch set whenever you push it | 05:08 |
*** m_kazuhiro has joined #openstack-swift | 05:16 | |
clayg | notmyname: i had it right the first time! | 05:23 |
openstackgerrit | Clay Gerrard proposed openstack/swift: Add POST capability to ssync for .meta files https://review.openstack.org/138498 | 05:31 |
*** mahatic_ has joined #openstack-swift | 05:31 | |
notmyname | wheeee | 05:31 |
clayg | my +2 is only sorta tenative - I'd like to get @acoles approval before it +A's if at all possible - I think he should be online soon? | 05:32 |
notmyname | yeah. 6:30am for him right now | 05:33 |
*** mahatic has quit IRC | 05:33 | |
clayg | great! ok, I'll pull down the next one :P | 05:37 |
clayg | i guess I should rebase it really | 05:37 |
clayg | oh god this one is bigger than the other one! | 05:38 |
notmyname | yeah, fixing POSTs was just to get ready for that one ;-) | 05:38 |
notmyname | however, most of the raw line count is from the tests, and they're pretty well-formatted IIRC | 05:39 |
notmyname | clayg: what do you mean by hoping nobody runs with post-as-copy=false? | 05:41 |
clayg | i forget - i do sorta remember saying that tho | 05:41 |
notmyname | yeah, I'm looking over your comments in the previous patch set | 05:42 |
clayg | oh, something about how we let object server's put datafiles under metadata files | 05:42 |
clayg | and metadata files only exist with wiht post-as-copy = false so it's only *barely* a behavior change because it's a change in the behavior of the non-default configuration | 05:42 |
openstackgerrit | Mahati Chamarthy proposed openstack/swift: Guess content type on PUT and probe tests https://review.openstack.org/220897 | 05:43 |
clayg | i mean it's only a backend behavior - it'll be fine | 05:43 |
clayg | ... probably | 05:43 |
clayg | this rebase is screwed :\ | 05:43 |
*** trifon has quit IRC | 05:43 | |
clayg | oh wait - i see it now | 05:44 |
*** trifon has joined #openstack-swift | 05:48 | |
clayg | shit - half the stuff I cleaned up in the last patch were stuff acoles fixed in *this* patch! | 05:48 |
clayg | bah w/e | 05:48 |
notmyname | bah! there was somethign else I was supposed to do tonight too. | 05:49 |
clayg | jok, i say "half" - i't more like two | 05:49 |
clayg | i used the old gerrit view just long enough that everything in the old view is in the wrong place now :'( | 05:51 |
openstackgerrit | Clay Gerrard proposed openstack/swift: Enable object server to return non-durable data https://review.openstack.org/215276 | 05:52 |
clayg | ... | 05:52 |
*** hrou has quit IRC | 06:01 | |
clayg | ok, so this didn't quite make it to X-Backend-If-Not-Frag-Timestamps - it seems like the proxy is still explicitly asking for things... | 06:03 |
clayg | instead of explicitly asking for anything but something that would be useless | 06:03 |
notmyname | clayg: this is what I get for the difference between acoles_'s patch sets and the 2 you just pushed https://gist.github.com/notmyname/c9cc9a9506430fe88343 | 06:06 |
clayg | I expect it to be like | 06:09 |
clayg | --- good acoles code | 06:10 |
clayg | +++ crappy clayg code | 06:10 |
clayg | did you just diff the sha's from the gerrit page? it's good that he like rebased recently I guess | 06:11 |
clayg | notmyname: but yeah that looks about right to me? | 06:11 |
notmyname | I had already pulled down the 2 patches locally into one branch. did the same after you pushed and diffed the 2 branches | 06:11 |
clayg | notmyname: ah - that's handy then | 06:12 |
notmyname | so if this looks sane to me, then I can look at the final without worry about the changes you've made and take them as part of the whole | 06:12 |
clayg | yeah i really don't think any of my changes were significant? I probably should have just done it all in follow up :\ | 06:12 |
notmyname | at least, I hope. this whole thing is rather mind-warping. I feel like I'm coming into all this, from a deep code level, late, so I've got catch-up to do :-/ | 06:13 |
clayg | i mean the docstring stuff was good | 06:13 |
clayg | i think pulling the 4 encode/decode methods out to module level functions will help with maintance and some initail grok | 06:13 |
notmyname | yeah. I've already been going back and forth between the 2 files to see each :-) | 06:14 |
clayg | fixing the TODO/meta stuff *would* have been significant it acoles hadn't already done in the non-durable patch | 06:14 |
notmyname | so yeah figuring out your 36 stream-of-consciousness review comments, though... ;-) | 06:15 |
clayg | totally useless right - you have to just ignore me - they're notes for me | 06:15 |
notmyname | (it's great. not a complaint) | 06:16 |
notmyname | they're good, because it helps understand why you were thinking stuff | 06:16 |
notmyname | do we really not have an overview doc on ssync? ie "here's what the protocol is" | 06:18 |
clayg | if you follow the eveloper docs through replicator and read the Sender and Reciver you can get quite a bit out of there | 06:18 |
notmyname | yeah, I know. I was hoping for the picture-book version ;-) | 06:19 |
clayg | NICE! | 06:19 |
*** mahatic_ has quit IRC | 06:20 | |
*** mahatic has joined #openstack-swift | 06:21 | |
*** dmorita has joined #openstack-swift | 06:22 | |
clayg | charz: notmyname: can we retrigger the cluster-swift-tox-func tests on patch 138498 | 06:25 |
patchbot | clayg: https://review.openstack.org/#/c/138498/ - Add POST capability to ssync for .meta files | 06:25 |
notmyname | clayg: I think a recheck comment will trigger that set of tests too | 06:26 |
charz | clayg: checking | 06:26 |
clayg | ah! very well then | 06:29 |
clayg | ok i'ma sign off for awhile | 06:29 |
clayg | acoles_: hopefully catch you in my am | 06:29 |
clayg | is tomorrow Friday already!? | 06:29 |
charz | clayg: the jobs is in queue. | 06:29 |
notmyname | hopefully that means until the morning ;-) | 06:29 |
clayg | notmyname: well see | 06:29 |
notmyname | yeah. already | 06:29 |
notmyname | I'll try to give this patch set a go and see where I get | 06:29 |
openstackgerrit | Mahati Chamarthy proposed openstack/swift: Guess content type on PUT and probe tests https://review.openstack.org/220897 | 06:30 |
*** mragupat has quit IRC | 06:30 | |
notmyname | ts. sometimes a tombstone. sometimes a timestamp. good luck | 06:36 |
notmyname | eg ts_meta and ts_data. surprise! this time it's a timestamp | 06:37 |
*** sanchitmalhotra has joined #openstack-swift | 06:43 | |
*** itlinux has joined #openstack-swift | 06:43 | |
*** sanchitmalhotra1 has quit IRC | 06:44 | |
*** sanchitmalhotra1 has joined #openstack-swift | 06:50 | |
*** sanchitmalhotra has quit IRC | 06:52 | |
*** wbhuber has joined #openstack-swift | 06:58 | |
*** mahatic has quit IRC | 07:01 | |
*** wbhuber has quit IRC | 07:02 | |
*** rledisez has joined #openstack-swift | 07:08 | |
*** tr_ has joined #openstack-swift | 07:29 | |
*** geaaru has joined #openstack-swift | 07:29 | |
*** chlong has quit IRC | 07:31 | |
*** kei_yama has quit IRC | 07:36 | |
*** itlinux has quit IRC | 07:38 | |
*** kei_yama has joined #openstack-swift | 07:40 | |
*** acoles_ is now known as acoles | 07:40 | |
acoles | clayg: yeah i'll around later | 07:41 |
*** itlinux has joined #openstack-swift | 07:41 | |
acoles | notmyname: its obvious , ts is a tombstamp :) | 07:42 |
*** haomaiwang has joined #openstack-swift | 07:44 | |
acoles | mattoliverau: thanks! | 07:45 |
*** openstackgerrit has quit IRC | 07:46 | |
*** openstackgerrit has joined #openstack-swift | 07:46 | |
*** jordanP has joined #openstack-swift | 07:47 | |
*** sanchitmalhotra has joined #openstack-swift | 07:51 | |
*** sanchitmalhotra1 has quit IRC | 07:53 | |
acoles | charz: thanks for doing a recheck on 138498. i'm interested to see if that passes on QA cluster before merging. | 07:54 |
acoles | notmyname: seriously, i had in mind to change ts to tomb where appropriate. obviously my mind is not the best place to make that kind of note. | 07:56 |
*** haomaiwang has quit IRC | 08:01 | |
*** haomaiwa_ has joined #openstack-swift | 08:01 | |
*** eranrom has joined #openstack-swift | 08:09 | |
*** kei_yama has quit IRC | 08:18 | |
*** kei_yama has joined #openstack-swift | 08:18 | |
*** jistr has joined #openstack-swift | 08:24 | |
*** peterlisak has joined #openstack-swift | 08:25 | |
*** tr_ has quit IRC | 08:30 | |
*** tab_ has joined #openstack-swift | 08:30 | |
*** cazino has joined #openstack-swift | 08:32 | |
acoles | charz: yay it passed this time | 08:35 |
charz | acoles: COOL!!! | 08:36 |
clayg | ohai! | 08:37 |
acoles | clayg: THANKS for all the reviewing! I am just going through, looks like we have ssync fix to ready land :D | 08:40 |
acoles | charz: yay for the QA cluster :) | 08:40 |
charz | LOL | 08:40 |
acoles | clayg: oh, sorry :/ i left a comment somewhere that those TODO's were fixed in the next patch, must have got hidden by new versions on gerrit | 08:43 |
clayg | s'ok this way they merge even if we can't the fix-all-the-things-proxy patch in | 08:44 |
clayg | acoles: I am sorta sniffing around the idea that now that we know *roughly* what the finished product will look like - we could try to split up the non-durable from the alt-frags change without too much worry that doing the first half without the second will paint us into a corner? | 08:46 |
clayg | but reading over your diff, there's still a number of TODO's in the grand-unified-fix-everything-we-could-possibly-want-to-fix-in-the-proxy patch | 08:46 |
clayg | notmyname probably won't cut anything release looking until Monday, but realistically today is probably the day - not sure what you've got already on your plate with the service token stuff? | 08:47 |
acoles | sure, its good to have those meta+tomb anomalies fixed. | 08:48 |
clayg | I also just remembered patch 181407 which ... would be very nice to have | 08:48 |
patchbot | clayg: https://review.openstack.org/#/c/181407/ - Don't ssync fragments that conflict with another f... | 08:48 |
acoles | clayg: service token fix is merged | 08:49 |
clayg | whooo hoo! | 08:49 |
acoles | kerching | 08:49 |
clayg | patch 227133 wasn't crossed off on the wiki | 08:49 |
patchbot | clayg: https://review.openstack.org/#/c/227133/ - Fix copy requests to service accounts in Keystone | 08:49 |
acoles | so that leaves all day to fix the whole world | 08:49 |
clayg | notmyname: cc^ slacking | 08:49 |
acoles | i'll cross it off | 08:49 |
*** links has quit IRC | 08:50 | |
clayg | a'ight well - i'm not going to be able to get a proper review of patch 215276 in tonite | 08:50 |
patchbot | clayg: https://review.openstack.org/#/c/215276/ - Enable object server to return non-durable data | 08:50 |
acoles | k so i can think about how dropping the alt frags may simplify 215276 patch and i will look at 181407 again | 08:51 |
acoles | clayg: u go sleep, i will still be here when you start your friday | 08:51 |
clayg | shit patch 222799 has a unittest failure :\ | 08:52 |
patchbot | clayg: https://review.openstack.org/#/c/222799/ - Validate against duplicate device part replica ass... | 08:52 |
*** aix has joined #openstack-swift | 08:54 | |
acoles | clayg: do i need to review that ^^ or do we have eyes on it? have to confess, the ring would be a learning curve for me. | 08:54 |
*** m_kazuhiro has quit IRC | 09:00 | |
*** haomaiwa_ has quit IRC | 09:01 | |
clayg | acoles: torgomatic has the ball | 09:01 |
*** haomaiwang has joined #openstack-swift | 09:01 | |
clayg | acoles: and cschwede maybe? | 09:01 |
clayg | acoles: mattoliverau looked at it | 09:01 |
clayg | acoles: I'd prefer if you could stay on the proxy/diskfile stuff | 09:02 |
acoles | clayg: oh phew. i love team effort. | 09:02 |
acoles | yeah not yielding hash when only a .meta makes a lot of sense. good call. | 09:02 |
openstackgerrit | Clay Gerrard proposed openstack/swift: Validate against duplicate device part replica assignment https://review.openstack.org/222799 | 09:02 |
clayg | who else is up? | 09:03 |
clayg | kota_: mattoliverau: ??? | 09:03 |
acoles | that would be a bit of ahospital pass | 09:03 |
acoles | yielding the meta file only that is | 09:04 |
clayg | google didn't know what ahospital was either :\ | 09:06 |
*** [1]eranrom has joined #openstack-swift | 09:09 | |
*** eranrom has quit IRC | 09:09 | |
*** [1]eranrom is now known as eranrom | 09:09 | |
*** tdasilva has quit IRC | 09:10 | |
clayg | hah! screw all you write-leader-electing-concensus-getting-consistent-promising distributed systems => http://danluu.com/limplock/ | 09:10 |
*** Guest64825 has quit IRC | 09:14 | |
*** dmsimard has quit IRC | 09:14 | |
*** dmorita has quit IRC | 09:14 | |
*** openstack has joined #openstack-swift | 09:23 | |
*** amandap has joined #openstack-swift | 09:23 | |
*** openstackstatus has joined #openstack-swift | 09:23 | |
*** ChanServ sets mode: +v openstackstatus | 09:23 | |
*** marzif has joined #openstack-swift | 09:29 | |
*** tdasilva has joined #openstack-swift | 09:32 | |
*** ChanServ sets mode: +v tdasilva | 09:32 | |
*** mahatic has joined #openstack-swift | 09:33 | |
*** cazino has left #openstack-swift | 09:34 | |
*** mahatic has quit IRC | 09:36 | |
*** marzif has quit IRC | 09:37 | |
*** marzif has joined #openstack-swift | 09:37 | |
*** haomaiwang has quit IRC | 10:01 | |
*** haomaiwang has joined #openstack-swift | 10:01 | |
*** daemontool__ has joined #openstack-swift | 10:06 | |
*** daemontool_ has quit IRC | 10:07 | |
*** [1]eranrom has joined #openstack-swift | 10:09 | |
*** eranrom has quit IRC | 10:09 | |
*** [1]eranrom is now known as eranrom | 10:09 | |
openstackgerrit | Peter Lisák proposed openstack/swift: swift-init return codes https://review.openstack.org/230352 | 10:09 |
*** pcaruana has joined #openstack-swift | 10:13 | |
*** cazino has joined #openstack-swift | 10:18 | |
*** sanchitmalhotra1 has joined #openstack-swift | 10:31 | |
*** sanchitmalhotra has quit IRC | 10:33 | |
*** mahatic has joined #openstack-swift | 10:38 | |
mahatic | acoles: hello | 10:39 |
mahatic | acoles: re: probe tests comment - won't unit tests need to skip too wrt crypto on/off? | 10:47 |
mahatic | acoles: ok, i don't think they do. My other question would be - are we not going to do any probe tests? there was a card in trello - which lead me to start these | 10:53 |
*** ig0r_ has joined #openstack-swift | 10:55 | |
acoles | mahatic: yeah, sorry, i only just realised that some of the testing could perhasp be done using the unit test pattern in test_server. IDK if there may still be some that need probe tests. | 10:55 |
*** ig0r_ has quit IRC | 10:56 | |
acoles | clayg: you didn't add yourself as co-author on patch 138498 | 10:59 |
patchbot | acoles: https://review.openstack.org/#/c/138498/ - Add POST capability to ssync for .meta files | 10:59 |
*** haomaiwang has quit IRC | 11:01 | |
mahatic | acoles: no problem. Also, you did trace it right in obj controller file - so if i wrap it and put it in request_helpers, you mean we could use that in future as well? also change it in controller/obj.py? | 11:01 |
*** haomaiwang has joined #openstack-swift | 11:01 | |
acoles | mahatic: i was thinking one function in request_helpers that's then called from obj.py and encrypter | 11:02 |
acoles | mahatic: but I didn't think about it for too long, so please double check my logic! | 11:02 |
acoles | mahatic: sorry i didn't have much time to review it today. | 11:02 |
*** cazino has quit IRC | 11:05 | |
mahatic | acoles: sure, will look into that. thank you! | 11:06 |
mahatic | acoles: and np, thanks for the review - I didn't expect it being friday today (for L) you'd look at it :) | 11:08 |
*** eranrom has quit IRC | 11:08 | |
*** eranrom has joined #openstack-swift | 11:09 | |
acoles | mahatic: sadly friday is included in the working week ;P have a good weekend | 11:15 |
*** delatte has joined #openstack-swift | 11:16 | |
*** cdelatte has joined #openstack-swift | 11:16 | |
mahatic | acoles: I know ;). Thanks, you have a good weekend too | 11:16 |
*** cazino has joined #openstack-swift | 11:24 | |
openstackgerrit | Alistair Coles proposed openstack/swift: Add POST capability to ssync for .meta files https://review.openstack.org/138498 | 11:24 |
*** ccavanna has quit IRC | 11:25 | |
acoles | clayg: nm i implicated you. great improvements, thanks. | 11:26 |
*** pcaruana has quit IRC | 11:47 | |
*** wbhuber has joined #openstack-swift | 11:52 | |
*** wbhuber has quit IRC | 11:54 | |
*** haomaiwang has quit IRC | 12:01 | |
*** haomaiwang has joined #openstack-swift | 12:01 | |
*** zul has quit IRC | 12:01 | |
*** zul has joined #openstack-swift | 12:04 | |
*** cazino has quit IRC | 12:05 | |
*** cazino has joined #openstack-swift | 12:05 | |
*** [1]eranrom has joined #openstack-swift | 12:09 | |
*** eranrom has quit IRC | 12:09 | |
*** [1]eranrom is now known as eranrom | 12:09 | |
*** haomaiwang has quit IRC | 12:10 | |
*** 20WAB697A has joined #openstack-swift | 12:11 | |
*** 20WAB697A has quit IRC | 12:25 | |
*** thumpba has joined #openstack-swift | 12:34 | |
*** delatte has quit IRC | 12:36 | |
*** cdelatte has quit IRC | 12:36 | |
*** delatte has joined #openstack-swift | 12:37 | |
*** cdelatte has joined #openstack-swift | 12:37 | |
*** MVenesio has joined #openstack-swift | 12:43 | |
*** MVenesio_ has joined #openstack-swift | 12:47 | |
*** thumpba_ has joined #openstack-swift | 12:48 | |
*** MVenesio has quit IRC | 12:50 | |
*** thumpba has quit IRC | 12:52 | |
*** MVenesio_ has quit IRC | 12:54 | |
*** MVenesio has joined #openstack-swift | 12:55 | |
*** dustins has joined #openstack-swift | 12:57 | |
*** proteusguy_ has joined #openstack-swift | 13:01 | |
*** [1]eranrom has joined #openstack-swift | 13:09 | |
*** eranrom has quit IRC | 13:09 | |
*** [1]eranrom is now known as eranrom | 13:09 | |
*** peterlisak has quit IRC | 13:10 | |
*** itlinux has quit IRC | 13:15 | |
*** itlinux has joined #openstack-swift | 13:15 | |
*** jrichli has joined #openstack-swift | 13:20 | |
*** tab_ has quit IRC | 13:21 | |
*** CR7 has joined #openstack-swift | 13:25 | |
*** haomaiwa_ has joined #openstack-swift | 13:25 | |
*** CR7 has quit IRC | 13:25 | |
*** hrou has joined #openstack-swift | 13:31 | |
*** cazino has quit IRC | 13:35 | |
*** cazino has joined #openstack-swift | 13:36 | |
*** haomaiwa_ has quit IRC | 13:37 | |
*** haomaiwang has joined #openstack-swift | 13:38 | |
*** haomaiwang has quit IRC | 13:43 | |
*** haomaiwang has joined #openstack-swift | 13:44 | |
*** itlinux has quit IRC | 13:50 | |
openstackgerrit | Merged openstack/swift: Add POST capability to ssync for .meta files https://review.openstack.org/138498 | 13:54 |
*** alejandrito has joined #openstack-swift | 13:55 | |
*** kei_yama has quit IRC | 13:59 | |
*** haomaiwang has quit IRC | 14:01 | |
*** lcurtis has joined #openstack-swift | 14:03 | |
*** haomaiwa_ has joined #openstack-swift | 14:04 | |
*** lcurtis has quit IRC | 14:04 | |
*** lcurtis has joined #openstack-swift | 14:04 | |
*** pberis has joined #openstack-swift | 14:06 | |
*** thurloat_isgone is now known as thurloat | 14:07 | |
*** thurloat is now known as thurloat_isgone | 14:09 | |
*** mahatic has quit IRC | 14:10 | |
*** haomaiwa_ has quit IRC | 14:12 | |
*** elmo_ is now known as elmo | 14:12 | |
*** haomaiwa_ has joined #openstack-swift | 14:13 | |
*** eranrom has quit IRC | 14:13 | |
*** thumpba_ has quit IRC | 14:14 | |
*** cazino has quit IRC | 14:23 | |
*** jbonjean has joined #openstack-swift | 14:26 | |
*** trifon has quit IRC | 14:27 | |
*** jlhinson has joined #openstack-swift | 14:31 | |
*** haomaiwa_ has quit IRC | 14:33 | |
*** wbhuber has joined #openstack-swift | 14:35 | |
*** haomaiwang has joined #openstack-swift | 14:35 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 14:36 |
*** cazino has joined #openstack-swift | 14:36 | |
*** haomaiwang has quit IRC | 14:40 | |
notmyname | good morning | 14:43 |
acoles | notmyname: hi | 14:45 |
notmyname | https://wiki.openstack.org/wiki/Swift/PriorityReviews updated :-) | 14:45 |
acoles | notmyname: 2 patches merged today: service token fix and ssync post fix | 14:46 |
notmyname | yup :-) | 14:46 |
notmyname | acoles: thanks for those | 14:46 |
notmyname | where I left patch 215276 last night was agreeing with clayg's comments and realizing that there's another patch | 14:47 |
patchbot | notmyname: https://review.openstack.org/#/c/215276/ - Enable object server to return non-durable data | 14:47 |
acoles | no, thanks you and clayg for a late night reviewing it seems | 14:47 |
acoles | another? | 14:47 |
notmyname | maybe? looking at it now, it doesn't seem I got too far on that one :-) | 14:50 |
notmyname | so ignore what I just said | 14:50 |
acoles | notmyname: clayg mentioned patch 181407 to me earlier so i have been breathing life back into that | 14:52 |
patchbot | acoles: https://review.openstack.org/#/c/181407/ - Don't ssync fragments that conflict with another f... | 14:52 |
acoles | notmyname: so far so good and its not that huge | 14:52 |
acoles | notmyname: but is based on 215276 in order to take advantage of the diskfile changes in there | 14:53 |
acoles | so i am wondering whether to split out some of the diskfile stuff from 215276 into a common base patch. and then there is clayg's comment about reducing scope of 215276 to just the non-durable getting | 14:54 |
acoles | i've not really had chance to consider that too much | 14:55 |
*** minwoob has joined #openstack-swift | 15:00 | |
*** Fin1te has joined #openstack-swift | 15:09 | |
*** chlong has joined #openstack-swift | 15:10 | |
clayg | acoles: I was hoping you wouldn't notice :\ | 15:14 |
acoles | clayg: notice what? | 15:15 |
clayg | acoles: i wan't implicated | 15:16 |
*** trifon has joined #openstack-swift | 15:16 | |
acoles | clayg: oh right. heh | 15:16 |
notmyname | oh, we all noticed ;-) | 15:16 |
acoles | clayg: i feel i made a real *hash* of that patch :D | 15:17 |
notmyname | acoles: which "part"? we should set up a timestone | 15:18 |
acoles | lol | 15:20 |
acoles | notmyname: as i heard said just this week, 'this code was hard to write, it should be hard to understand' :) | 15:21 |
clayg | acoles: ^ that's a redbo-ism! | 15:23 |
acoles | well redbo is famous worldwide then | 15:24 |
clayg | ok, I need to get the kids ready for school | 15:24 |
clayg | i'll rebase patch 215276 and probably try to split the .durable from the alt-frags stuff | 15:24 |
patchbot | clayg: https://review.openstack.org/#/c/215276/ - Enable object server to return non-durable data | 15:24 |
*** lyrrad has joined #openstack-swift | 15:24 | |
clayg | we may have to cut the rc w/o any of it? | 15:24 |
acoles | notmyname: clayg is it worth doing a hangout to discuss where we go with 215276? | 15:25 |
acoles | later that is | 15:25 |
notmyname | yeah, I think so | 15:25 |
clayg | if next week while testing I feel like I can make a case that we *need* one or both - i'll have the split up | 15:25 |
clayg | if you're still on in a few hours - i need to get off for now | 15:25 |
acoles | k | 15:26 |
clayg | few/couple | 15:26 |
acoles | i'll be here fo a bit longer | 15:26 |
clayg | you can also type up some stuff | 15:26 |
clayg | afk | 15:26 |
acoles | i think i'll have patch 181407 in good shape | 15:26 |
patchbot | acoles: https://review.openstack.org/#/c/181407/ - Don't ssync fragments that conflict with another f... | 15:26 |
*** chsc has joined #openstack-swift | 15:28 | |
notmyname | I need to get ready too | 15:29 |
timburke | good morning | 15:35 |
*** cazino has quit IRC | 15:35 | |
*** klrmn1 has joined #openstack-swift | 15:36 | |
*** thumpba has joined #openstack-swift | 15:38 | |
*** thumpba has quit IRC | 15:38 | |
*** nadeem has joined #openstack-swift | 15:39 | |
*** thumpba has joined #openstack-swift | 15:39 | |
*** nadeem has quit IRC | 15:40 | |
*** nadeem has joined #openstack-swift | 15:41 | |
*** janonymous_ has joined #openstack-swift | 15:44 | |
*** klrmn1 has quit IRC | 15:47 | |
*** cazino has joined #openstack-swift | 15:47 | |
*** silor has joined #openstack-swift | 15:52 | |
*** pberis has quit IRC | 15:59 | |
*** silor has quit IRC | 16:05 | |
*** jordanP has quit IRC | 16:09 | |
*** CR7 has joined #openstack-swift | 16:10 | |
*** jordanP has joined #openstack-swift | 16:11 | |
*** rledisez has quit IRC | 16:18 | |
clayg | acoles: you need to type yes or something on patch 181407 | 16:23 |
patchbot | clayg: https://review.openstack.org/#/c/181407/ - Don't ssync fragments that conflict with another f... | 16:23 |
clayg | acoles: idk how it's gunna work out - I have a commute followed by an eng meeting - i'm probably still ~2hrs from skyping | 16:24 |
acoles | clayg: as in yes i am working on it? | 16:25 |
acoles | clayg: k, well i may commute home and can hangout later | 16:25 |
clayg | acoles: ! | 16:25 |
*** pberis has joined #openstack-swift | 16:25 | |
acoles | clayg: where do i need to type yes? | 16:25 |
clayg | acoles: I saw "i think i have patch 181407 in good shape" | 16:25 |
patchbot | clayg: https://review.openstack.org/#/c/181407/ - Don't ssync fragments that conflict with another f... | 16:25 |
clayg | and when I saw it was still in merge conflict I thought you forgot to push | 16:26 |
clayg | but you're just being optomistic about the future - THAT'S GOOD! | 16:26 |
clayg | ok, i'm heading in - afk again | 16:26 |
acoles | clayg: oic. yes! just putting a probe test in | 16:26 |
clayg | acoles: hopefully talk to you later - but if not next week! | 16:26 |
clayg | l84 | 16:26 |
acoles | iirc you wrote the probe test | 16:26 |
*** cazino has left #openstack-swift | 16:27 | |
*** diogogmt has joined #openstack-swift | 16:32 | |
*** jordanP has quit IRC | 16:36 | |
*** pberis has quit IRC | 16:39 | |
*** pberis has joined #openstack-swift | 16:39 | |
*** wbhuber_ has joined #openstack-swift | 16:41 | |
*** janonymous_ has quit IRC | 16:47 | |
*** diogogmt_ has joined #openstack-swift | 16:48 | |
*** wbhuber_ has quit IRC | 16:48 | |
*** diogogmt has quit IRC | 16:50 | |
*** diogogmt_ has quit IRC | 16:53 | |
openstackgerrit | Alistair Coles proposed openstack/swift: Don't ssync fragments to handoffs if they conflict https://review.openstack.org/181407 | 16:56 |
*** jlhinson_ has joined #openstack-swift | 16:56 | |
openstackgerrit | Alistair Coles proposed openstack/swift: Enable object server to return non-durable data https://review.openstack.org/215276 | 16:56 |
*** Fin1te has quit IRC | 16:57 | |
acoles | clayg: i typed yes. 181407 depends on 215276 because it uses the DiskFile.fragments property, so we may end up wanting to rip some of the diskfile changes out of 216276 into a separate patch and base 181407 on that. | 16:57 |
*** jlhinson has quit IRC | 16:59 | |
acoles | clayg: and you are an author on 181407 because you wrote the probe test. i took it from patch 176403 | 16:59 |
patchbot | acoles: https://review.openstack.org/#/c/176403/ - WIP Revert to another handoff when primary is unav... | 16:59 |
acoles | ^^ that is probably now redundant | 16:59 |
*** diogogmt has joined #openstack-swift | 17:00 | |
*** Fin1te has joined #openstack-swift | 17:02 | |
*** jistr has quit IRC | 17:10 | |
*** geaaru has quit IRC | 17:11 | |
hrou | Hey timburke, have some time to chat about the encoding issue you pointed out in the review of patch 194191 ? I can just leave a comment in the review if you'd find that easier. | 17:11 |
patchbot | hrou: https://review.openstack.org/#/c/194191/ - Encrypter and Decrypter middleware - object body only | 17:11 |
*** jrichli has quit IRC | 17:12 | |
*** garthb has joined #openstack-swift | 17:13 | |
* notmyname back online | 17:14 | |
*** Fin1te has quit IRC | 17:16 | |
*** klrmn1 has joined #openstack-swift | 17:16 | |
timburke | hrou: sure! i've got a meeting in about 15min, so it might be better to wait until after that, though. | 17:17 |
hrou | timburke, awesome thanks! Yea lets wait till after you're meeting. | 17:18 |
timburke | but it couldn't hurt to talk, and give me something to mull over instead of paying attention to the meeting :P | 17:18 |
hrou | timburke, sure in that case : ), here goes: | 17:18 |
hrou | timburke, So firstly, thanks again for pointing it out, it will indeed cause problems with py3. | 17:18 |
hrou | timburke, Now, you've suggested we serialize the crpyto meta to a json formatted string, for the IV (which is a bytes object), we will first base64 encode it. | 17:19 |
hrou | timburke, just wanted to point out one minor issue with this (but it has an easy fix); json.dumps expects a string (unicode) not bytes (which is the output of encode), so a small fix to this is to take the encoded binary decode it in utf-8 (i.e. decode('utf-8') ). | 17:19 |
timburke | (or even the default, us-ascii. it sure *better* be ascii!) | 17:20 |
hrou | timburke, so all of the above will work in py2 and 3 just fine. But leads to my actual question, what's the advantage of JSON'ing the object, an alternative (which you hinted at too ; ) is to simply encode crypto_meta[iv], which is a bytes object in 3.0 (i.e. the encoding will work). | 17:21 |
hrou | timburke, haha hmm I thought the default was utf-8 was the default actually, I'll check. | 17:22 |
timburke | we've got structure that we're encoding, too, though. or are you thinking of using separate headers for iv and cipher? that'd be fine, too | 17:23 |
hrou | timburke, ah wait I see you're suggesting just leave it to the default which would naturally be utf-8 in py3, yep that makes sense ! | 17:23 |
hrou | timburke, well the debate is do we need to encode the entire structure ? | 17:24 |
hrou | timburke, the rest is 100% ascii, but maybe to future proof (name the crypto algo) its better to encode the entire thing. | 17:24 |
timburke | and maybe we don't. i wasn't sure how much else you might want to stuff in there, so was looking for ways to still just use the one header | 17:25 |
hrou | timburke, yea I think I'd rather leave it as one header for now, so maybe the encoding route is better. | 17:25 |
timburke | including the cipher definitely seems like a good idea. surely at some point we'll want the new AAES_CTR_1024 or some madness, and we'll want a way to distinguish them without relying on defaults | 17:28 |
hrou | Yep absolutely ! | 17:28 |
hrou | So I'm liking the proposal of json'ing the object and encoding the IV. | 17:28 |
*** garthb has quit IRC | 17:29 | |
*** daemontool__ has quit IRC | 17:29 | |
timburke | (imagine if we'd done this in the long long ago, and now had to deal with headers.get(cipher_header, 'DES_ECB') code... bleh!) | 17:29 |
timburke | cool. sounds good. | 17:29 |
timburke | see, 15mins was perfect! | 17:30 |
*** daemontool__ has joined #openstack-swift | 17:30 | |
hrou | haha yea that'd be pretty terrible. | 17:30 |
hrou | And that was no coincidence at all ; ) Thanks so much timburke ! It was neat ,got to learn more about py3 though this, which is always fun. | 17:31 |
timburke | yeah... i'm kinda dreading swift actually trying to support it. the fact that there's *still* unicode bugs in swiftclient doesn't make me feel good | 17:32 |
hrou | yea they kinda of jump out at you once you've dealt with one, ah interesting enough it seems they dealt with the issue I cited above in the nova client, patch 95623 | 17:33 |
patchbot | hrou: https://review.openstack.org/#/c/95623/ - In Py3 decode the output of base64.decode | 17:33 |
*** ChanServ has quit IRC | 17:37 | |
acoles | notmyname: clayg i am heading home, back on line later (maybe around noon your time?) | 17:38 |
*** ChanServ has joined #openstack-swift | 17:40 | |
*** cameron.freenode.net sets mode: +o ChanServ | 17:40 | |
*** acoles is now known as acoles_ | 17:41 | |
*** gyee has joined #openstack-swift | 17:50 | |
notmyname | acoles_: ok, great. we'd love to get on the phone when you have time | 17:57 |
*** delatte has quit IRC | 18:06 | |
*** cdelatte has quit IRC | 18:06 | |
*** nadeem has quit IRC | 18:06 | |
*** garthb has joined #openstack-swift | 18:06 | |
*** jrichli has joined #openstack-swift | 18:07 | |
*** janonymous_ has joined #openstack-swift | 18:10 | |
*** wbhuber has quit IRC | 18:18 | |
*** wbhuber has joined #openstack-swift | 18:18 | |
*** fbo has joined #openstack-swift | 18:21 | |
*** daemontool__ has quit IRC | 18:22 | |
*** pberis has quit IRC | 18:28 | |
*** CR7 has quit IRC | 18:32 | |
clayg | acoles_: shit you had to rebase patch 181407 on patch 215276 :'( | 18:40 |
patchbot | clayg: https://review.openstack.org/#/c/181407/ - Don't ssync fragments to handoffs if they conflict | 18:40 |
patchbot | clayg: https://review.openstack.org/#/c/215276/ - Enable object server to return non-durable data | 18:40 |
clayg | torgomatic: hey whenever you're looking at patch 222799 - if you need to high bandwidth anything just let me know? | 18:41 |
patchbot | clayg: https://review.openstack.org/#/c/222799/ - Validate against duplicate device part replica ass... | 18:41 |
*** eranrom has joined #openstack-swift | 18:46 | |
*** pberis has joined #openstack-swift | 18:53 | |
*** sanchitmalhotra has joined #openstack-swift | 18:57 | |
*** ccavanna has joined #openstack-swift | 18:59 | |
*** sanchitmalhotra1 has quit IRC | 18:59 | |
*** Fin1te has joined #openstack-swift | 19:01 | |
*** jrichli has quit IRC | 19:02 | |
*** Fin1te has quit IRC | 19:03 | |
*** marcusvrn_ has quit IRC | 19:05 | |
*** acoles_ is now known as acoles | 19:09 | |
*** wolsen has quit IRC | 19:10 | |
clayg | acoles: hey! | 19:13 |
acoles | clayg: hi, i'm back | 19:13 |
notmyname | great | 19:14 |
*** nadeem has joined #openstack-swift | 19:14 | |
notmyname | acoles: clayg: want to get on video chat and talk about these last patches? | 19:14 |
acoles | notmyname: sure | 19:14 |
*** daemontool_ has joined #openstack-swift | 19:14 | |
clayg | notmyname: sure! | 19:15 |
clayg | notmyname: but I don't know how :'( | 19:15 |
notmyname | I can. let me finish shoveling this food in my mouth | 19:15 |
clayg | nom nom nom | 19:15 |
notmyname | noms done | 19:16 |
acoles | i can wait if you guys are lunching | 19:16 |
notmyname | I'm done. and clayg absorbs nutrients from the air, I think | 19:16 |
clayg | the code - it feeds me | 19:17 |
*** nadeem has quit IRC | 19:17 | |
*** wolsen has joined #openstack-swift | 19:17 | |
*** nadeem has joined #openstack-swift | 19:17 | |
acoles | 'tis manna | 19:17 |
*** jrichli has joined #openstack-swift | 19:21 | |
*** subscope has joined #openstack-swift | 19:24 | |
*** subscope_ has joined #openstack-swift | 19:25 | |
*** marcusvrn_ has joined #openstack-swift | 19:26 | |
acoles | i just saw the t shirt | 19:26 |
*** CR7 has joined #openstack-swift | 19:27 | |
*** david-lyle has quit IRC | 19:35 | |
*** david-lyle has joined #openstack-swift | 19:35 | |
*** CR7 has quit IRC | 19:38 | |
jrichli | acoles: did I miss a link? I checked twitter ... nothing but the 'hint' ... | 19:39 |
acoles | its on notmyname's body | 19:39 |
jrichli | oh! you guys are ... having dinner? | 19:40 |
*** [1]eranrom has joined #openstack-swift | 19:43 | |
*** eranrom has quit IRC | 19:43 | |
*** [1]eranrom is now known as eranrom | 19:43 | |
acoles | jrichli: skype | 19:44 |
*** david-ly_ has joined #openstack-swift | 19:45 | |
jrichli | oh right ... i guess i saw that you guys might be on a call later | 19:45 |
jrichli | so, how does it look??? | 19:46 |
jrichli | is there a bird? is it top secret? | 19:46 |
acoles | jrichli: i am sworn to secrecy | 19:46 |
jrichli | i figured ;-) | 19:46 |
acoles | it has short sleeves | 19:46 |
jrichli | gee, thx | 19:47 |
acoles | cotton i am guessing | 19:47 |
acoles | dark, but we knew that | 19:47 |
notmyname | lol | 19:47 |
notmyname | actually only partly cotton. it's a blend. super soft | 19:47 |
acoles | they could have chosen a better model :P :P | 19:48 |
notmyname | TRUE THAT! | 19:48 |
acoles | oooh super soft. well i couldn't actualy touch it. and that would have been weird anyway. | 19:48 |
*** david-lyle has quit IRC | 19:49 | |
*** david-ly_ has quit IRC | 19:49 | |
jrichli | maybe i could find a source from swiftstack who hasn't been sworn to secrecy yet ... | 19:49 |
acoles | clayg: oh i forgot to say that i have a follow on for 215276 that removes the durable requirement completely. but, later... | 19:49 |
acoles | jrichli: actually i'm thinking the whole thing is a decoy | 19:50 |
jrichli | I actually pictured a shirt with a big duck on it | 19:50 |
acoles | jrichli: it did kinda look like pieces of sticky tape on a plain tee | 19:50 |
jrichli | but i know what u mean ;-) | 19:51 |
notmyname | they were packed up and shipping to japan today. I just grabbed one early :-) | 19:51 |
acoles | jrichli: quick, get to the port and intercept... | 19:52 |
notmyname | ok, so acoles and clayg and I just chatted about the outstanding patches. for those who are curious, here's the plan: | 19:52 |
notmyname | we need to land patch 222799 today | 19:52 |
patchbot | notmyname: https://review.openstack.org/#/c/222799/ - Validate against duplicate device part replica ass... | 19:52 |
acoles | i'm curious ;) | 19:52 |
*** thumpba has quit IRC | 19:53 | |
notmyname | and python 215276 will not land today. we'll keep working on it, though | 19:53 |
acoles | python!! | 19:53 |
notmyname | instead, clayg is working on a new simple patch to increase the quorum for durables | 19:53 |
notmyname | that combined with the other changes that have gone in recently should seriously mitigate any concerns with but 1452431 | 19:54 |
notmyname | bug 1452431 | 19:54 |
openstack | bug 1452431 in OpenStack Object Storage (swift) "some parts replicas assigned to duplicate devices in the ring" [Critical,Confirmed] https://launchpad.net/bugs/1452431 - Assigned to Samuel Merritt (torgomatic) | 19:54 |
notmyname | although, the bug isn't completely closed now, the likelyhood is much much less, we suspect, of happening | 19:54 |
notmyname | so with clayg's new patch and patch 222799 and my patch for the authors/changelog update, we'll have our liberty release | 19:54 |
patchbot | notmyname: https://review.openstack.org/#/c/222799/ - Validate against duplicate device part replica ass... | 19:54 |
notmyname | and at leat the 3 of us feel pretty much ok with it :-) | 19:55 |
*** CR7 has joined #openstack-swift | 19:55 | |
eranrom | I have been looking at it for the past few days, and confess I am only starting to figure out that builder code. | 19:56 |
eranrom | So I take it is no longer required (FWIW) | 19:56 |
eranrom | I am referring to https://review.openstack.org/#/c/222799/ | 19:56 |
acoles | eranrom: that IS required | 19:57 |
acoles | eranrom: patch 215276 is deferred | 19:57 |
patchbot | acoles: https://review.openstack.org/#/c/215276/ - Enable object server to return non-durable data | 19:57 |
eranrom | acoles: So should I coninue reviewing the duplicate devices then? | 19:58 |
eranrom | again FWIW | 19:58 |
*** thumpba has joined #openstack-swift | 19:59 | |
acoles | eranrom: i was just clarifying that 222799 is still in scope for the release, notmyname: clayg should answer if they need your review | 19:59 |
*** mac_ified has joined #openstack-swift | 19:59 | |
eranrom | ok thanks! | 20:00 |
*** acoles is now known as acoles_ | 20:02 | |
notmyname | yeah, any reviews on patch 222799 or the one clayg will push up later will be very helpful | 20:05 |
patchbot | notmyname: https://review.openstack.org/#/c/222799/ - Validate against duplicate device part replica ass... | 20:05 |
eranrom | ok, sure | 20:07 |
notmyname | I just saw this article about how swift is used to help make movies/tv shows. http://dailycloud.info/somewhere-over-the-rainbow-cloud-instances-shine/ | 20:09 |
*** CR7 has quit IRC | 20:11 | |
janonymous_ | link is awesome : http://lmgtfy.com/?q=pigeonhole+principle :P | 20:12 |
*** CR7 has joined #openstack-swift | 20:17 | |
*** thumpba has quit IRC | 20:17 | |
*** pberis has quit IRC | 20:18 | |
*** david-lyle has joined #openstack-swift | 20:24 | |
*** Fin1te has joined #openstack-swift | 20:25 | |
*** CR7 has quit IRC | 20:30 | |
*** CR7 has joined #openstack-swift | 20:30 | |
*** Fin1te has quit IRC | 20:36 | |
*** [1]eranrom has joined #openstack-swift | 20:43 | |
*** eranrom has quit IRC | 20:43 | |
*** [1]eranrom is now known as eranrom | 20:43 | |
clayg | oh yeah, hrmm... I sorta expected I might get dinged for that in the review being as that's it's *kinda* snarky - but oh well - maybe we're all cool with a having a little bit of fun with stuff like that? | 20:44 |
clayg | notmyname: wow - *one* test failed the durable quorum change | 20:44 |
notmyname | not bad! | 20:45 |
notmyname | or terrifying | 20:45 |
janonymous_ | clayg : It's cool :) | 20:47 |
janonymous_ | I was having a look at swift vids and i saw this: https://www.openstack.org/summit/vancouver-2015/summit-videos/presentation/what-and-039s-up-swift-the-globally-distributed-community-behind-swift :) | 20:50 |
clayg | janonymous_: man - just *look* at those handsome blokes! | 20:52 |
clayg | we were so young | 20:52 |
notmyname | EC has aged us | 20:54 |
janonymous_ | clayg: Hahaa. It was a nice talk :) | 20:54 |
*** ccavanna has quit IRC | 20:55 | |
torgomatic | Some people age like fine wines; I age more like leftover pizza. | 20:57 |
clayg | we *still* have that stupid ChunkWriteTimeout popping up in tests? rly? | 20:58 |
clayg | I'm about to go stuck a calling frame on that fucker and find out who made him | 20:59 |
openstackgerrit | Clay Gerrard proposed openstack/swift: Make sure we have enough .durable's for GETs https://review.openstack.org/230641 | 20:59 |
clayg | you know... when you type out a description of the change like that it sounds *entirely* too reasonable :\ | 21:00 |
notmyname | yup. looks very reasonable when looking at the diff | 21:03 |
*** dustins has quit IRC | 21:03 | |
notmyname | after I put down the other patch, I'll run tests on this one and leave a reveiw | 21:03 |
clayg | notmyname: you may wait and see if the probetests will pass :\ | 21:04 |
notmyname | anyone have in-progress reviews on python 222799? I wanted to update the commit message | 21:04 |
notmyname | clayg: of course. I'll run the tests :-) | 21:04 |
clayg | i don't *think* probetests really do anything to much with timeouts on .durables tho | 21:05 |
clayg | and we require the same number of 201's we always did *really* (at least for the frag archives themselves) | 21:05 |
clayg | a'ight ChunkWriteTimeout - let's see if you can hide *now* | 21:05 |
*** MVenesio has quit IRC | 21:06 | |
notmyname | yeah. probably a place for some additional probe tests | 21:06 |
clayg | next step is to hack nosetest runner to sleep 60 before exiting and doing a binary search module by module | 21:06 |
clayg | well that or a sysexit handler in eventlet that will wait for the hub to clear it's timeouts... | 21:07 |
clayg | it's gotta be in a spawn_n somewhere | 21:07 |
clayg | File "/vagrant/swift/swift/proxy/controllers/obj.py", line 1407, in put_fragments_in_queue\n for fragment in frag_iter:\n', ' File "/vagrant/swift/swift/proxy/controllers/base.py", line 881, in iter_bytes_from_response_part\n self.app.client_timeout | 21:08 |
notmyname | didn't I just add some spawn_ns to swift? | 21:08 |
clayg | figues - the stack is totally disconnected from the test that spawned it (stupid greenlets) | 21:08 |
clayg | notmyname: i don't think so - you had regular spawns? | 21:09 |
notmyname | hmm..yeah. earlier versions used spawn_n | 21:09 |
notmyname | I think | 21:09 |
notmyname | ..back to rings | 21:09 |
notmyname | the comments in test_warn_at_risk() and test_no_warn_when_balanced are really helpful! | 21:10 |
*** thumpba has joined #openstack-swift | 21:10 | |
clayg | that seems so unlikely... | 21:10 |
clayg | oh those | 21:11 |
clayg | yeah kota tried to add those comments and like - gah - it's really hard to explain | 21:11 |
*** wbhuber has quit IRC | 21:13 | |
*** trifon has quit IRC | 21:17 | |
*** jrichli has quit IRC | 21:18 | |
clayg | ... this may not work - but I've attached the getter that's firing the timeout to the timeout so in addition to the stack frame I can include some of the getter's attributes like req_path, and statuses - see if it gives me a hint which test is starting up this thing | 21:20 |
*** diogogmt_ has joined #openstack-swift | 21:26 | |
*** diogogmt has quit IRC | 21:29 | |
*** diogogmt_ is now known as diogogmt | 21:29 | |
*** nadeem has quit IRC | 21:35 | |
*** jlhinson has joined #openstack-swift | 21:39 | |
*** jlhinson_ has quit IRC | 21:39 | |
*** [1]eranrom has joined #openstack-swift | 21:44 | |
*** eranrom has quit IRC | 21:44 | |
*** [1]eranrom is now known as eranrom | 21:44 | |
*** marcusvrn_ has quit IRC | 21:45 | |
*** thumpba has quit IRC | 21:49 | |
*** jlhinson has quit IRC | 21:50 | |
*** janonymous_ has quit IRC | 21:51 | |
*** pgbridge has quit IRC | 21:55 | |
*** subscope_ has quit IRC | 21:56 | |
*** pgbridge has joined #openstack-swift | 21:59 | |
*** CaioBrentano has quit IRC | 22:01 | |
notmyname | so set the subtier to the smaller of the roomiest tier or fudgiest tier, then figure out the sibling tiers and put it in one that is unused. set the tier to the subtier and loop | 22:03 |
notmyname | it's all so clear now | 22:03 |
*** hrou has quit IRC | 22:06 | |
*** minwoob_ has joined #openstack-swift | 22:12 | |
eranrom | notmyname: I ask myself if there is a point in looking at non direct siblings in this case. | 22:13 |
*** minwoob has quit IRC | 22:13 | |
notmyname | eranrom: isn't it like 1am for you? | 22:14 |
eranrom | that is in the case where all siblings already hold a replica of the partition | 22:14 |
eranrom | it is :-) | 22:14 |
clayg | alright! now I have some details /v1/a/c/o - so *that* narrows it down | 22:15 |
clayg | eranrom: back tracking would be an improvement; but not going down the wrong path would be a solution | 22:16 |
clayg | eranrom: the *real* weeds start to get pulled up when you realize your problem isn't in _reassign_parts it's in *gather* parts - then you say f' it - I'll ship this and loop back around next week ;) | 22:17 |
clayg | 'Range': 'bytes=0-1927' < grep for 1927 didn't turn up anything | 22:18 |
eranrom | clayg, well, I still did not get to that (gather) part yet in my exploration of the code... | 22:19 |
clayg | eranrom: heh - i'm just funnin' - I don't think that change acctually touches gather | 22:20 |
eranrom | it does not | 22:20 |
notmyname | clayg: I thought you had added a check that you couldn't actually have a ring at all with less devices than parts | 22:21 |
openstackgerrit | Zack M. Davis proposed openstack/swift: replace use of deprecated rfc822.Message with a helper utility https://review.openstack.org/203304 | 22:21 |
clayg | notmyname: turned it into a warning | 22:21 |
notmyname | ok | 22:21 |
notmyname | and there are *lots* of warnings | 22:21 |
clayg | notmyname: since we're not going to restrict double subscribing devices to different replicas of the same part there didn't seem like a strong reason to make it blow up hard | 22:22 |
clayg | notmyname: I think that one *could* be made an error tho - since we allow set_replicas - and reducing the replica count is cheaper than growing new replicas of parts | 22:22 |
eranrom | I will post a short review comment on what I have so far (haven't gone through the test/unit/common/ring/test_builder.py) will continue tomorrow assuming it will still be relevant... | 22:24 |
notmyname | eranrom: perhaps, but unlikely. I'm hoping to get this landed tonight | 22:25 |
notmyname | I'm also hoping that torgomatic is looking at it | 22:25 |
eranrom | notmyname,ok, then hopefully what I have is helpful in any way :-) | 22:26 |
notmyname | eranrom: thanks | 22:26 |
*** lcurtis has quit IRC | 22:26 | |
*** subscope has quit IRC | 22:30 | |
clayg | eranrom: yeah I'm sure it'll be great! | 22:30 |
*** chsc has quit IRC | 22:33 | |
*** eranrom has quit IRC | 22:39 | |
clayg | hrmm.... it might be an interesting fallback to pick just *any* other devices w/ort tier - basically throw out dispersion and balance in favor of not doing something totally stupid - seems like a cludge - picking another device on the same server is mostly benign | 22:43 |
*** david-lyle has quit IRC | 22:45 | |
*** CR7 has quit IRC | 22:46 | |
*** david-lyle has joined #openstack-swift | 22:47 | |
clayg | so my nose trick didn't work - but I'm not sure if it's because the gt that spawned the thing was closed and gc shut something down before my sleep - or if it was just cause my sleep wasn't green so the hub wasn't able to spin? | 22:49 |
*** ajiang has quit IRC | 22:49 | |
clayg | ... either way I pulled it out and I'm trying to wait in test.unit.proxy.test_server.teardown | 22:49 |
*** marzif has quit IRC | 22:49 | |
clayg | ... I'm pretty sure it's in proxy.test_server | 22:49 |
*** bhakta has quit IRC | 22:49 | |
*** itlinux has joined #openstack-swift | 22:53 | |
clayg | shit - i think i added it with the ec-disconn tests :\ | 22:57 |
*** itlinux has quit IRC | 22:57 | |
notmyname | I think half the problem with understanding ring patches is the variable naming and nested data structures | 23:00 |
clayg | notmyname: YES! | 23:01 |
clayg | i was *thinking* about that! | 23:01 |
notmyname | what's tier in this context? what's that nested list data structure look like? whoa! here comes a tuple! | 23:01 |
clayg | I think ... well i have some thoughts | 23:01 |
notmyname | and why are we talking about fudge? and where can I get some? | 23:02 |
notmyname | clayg: I suspect you may have written all the "fudgy" stuff after visiting your grandmother for the holidays or something | 23:03 |
notmyname | here's the fudge tiers. over there are the pralines. here's the pecan pie | 23:03 |
*** gyee has quit IRC | 23:04 | |
*** alejandrito has quit IRC | 23:10 | |
clayg | notmyname: no i didn't write any of that! | 23:11 |
notmyname | heh | 23:11 |
*** wbhuber has joined #openstack-swift | 23:12 | |
clayg | i'm mostly free of git blame under swift.common.ring - well... maybe that's not true because of dispersion reports | 23:12 |
notmyname | yeah, looks like torgomatic | 23:12 |
*** itlinux has joined #openstack-swift | 23:14 | |
notmyname | I wonder if mattoliverau is doing anything this saturday am | 23:15 |
*** MVenesio has joined #openstack-swift | 23:17 | |
*** MVenesio has quit IRC | 23:18 | |
*** itlinux has quit IRC | 23:19 | |
mattoliverau | notmyname: I'm brunching, but not leaving for about 1/2 hour.. Want me to attempt to look at the ring change again? | 23:21 |
notmyname | it would be helpful | 23:21 |
notmyname | mattoliverau: also, https://review.openstack.org/#/c/230641/ is short and needs to land in lieu of patch 215276 | 23:22 |
patchbot | notmyname: https://review.openstack.org/#/c/215276/ - Enable object server to return non-durable data | 23:22 |
mattoliverau | Also, yes, fudge factor is, I'm sure useful, but a little confusing and not at all as delicious as it sounds | 23:22 |
mattoliverau | Hai | 23:22 |
*** garthb has quit IRC | 23:34 | |
*** nadeem has joined #openstack-swift | 23:41 | |
openstackgerrit | Clay Gerrard proposed openstack/swift: Validate against duplicate device part replica assignment https://review.openstack.org/222799 | 23:42 |
clayg | lol - no it's not tasty at all | 23:43 |
clayg | idk, maybe it would have been useful - what if it's not obvious why you can't have three replicas on two devices? | 23:47 |
notmyname | no | 23:49 |
notmyname | :-) | 23:49 |
*** diogogmt_ has joined #openstack-swift | 23:51 | |
*** diogogmt has quit IRC | 23:52 | |
*** diogogmt_ is now known as diogogmt | 23:52 | |
*** diogogmt has quit IRC | 23:54 | |
*** diogogmt has joined #openstack-swift | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!