| *** ho has joined #openstack-swift | 00:22 | |
| ho | good morning guys! | 00:23 |
|---|---|---|
| *** bill_az has joined #openstack-swift | 00:50 | |
| *** abhirc has quit IRC | 00:51 | |
| *** abhirc has joined #openstack-swift | 00:51 | |
| *** abhirc has quit IRC | 00:55 | |
| *** abhirc has joined #openstack-swift | 00:55 | |
| *** abhirc has quit IRC | 00:57 | |
| *** abhirc has joined #openstack-swift | 00:57 | |
| *** abhirc has quit IRC | 00:58 | |
| *** abhirc has joined #openstack-swift | 00:58 | |
| *** abhirc has quit IRC | 01:05 | |
| *** yuan has joined #openstack-swift | 01:24 | |
| *** bill_az has quit IRC | 01:37 | |
| *** abhirc has joined #openstack-swift | 01:40 | |
| *** haomaiwang has joined #openstack-swift | 02:26 | |
| *** Krast has joined #openstack-swift | 02:32 | |
| *** echevemaster has quit IRC | 02:54 | |
| *** sandywalsh has quit IRC | 03:45 | |
| *** kota_ has joined #openstack-swift | 04:35 | |
| *** ppai has joined #openstack-swift | 04:42 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:44 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:09 | |
| openstackgerrit | Prashanth Pai proposed openstack/swift: Make object creation more atomic in Linux https://review.openstack.org/162243 | 06:10 |
| *** kota_ has quit IRC | 06:11 | |
| *** kota_ has joined #openstack-swift | 06:14 | |
| *** ktsuyuzaki has quit IRC | 06:15 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:16 | |
| *** zhill has joined #openstack-swift | 06:18 | |
| *** kota_ has quit IRC | 06:18 | |
| *** kota_ has joined #openstack-swift | 06:19 | |
| *** ktsuyuzaki has quit IRC | 06:21 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:21 | |
| *** kota_ has quit IRC | 06:23 | |
| *** kota_ has joined #openstack-swift | 06:24 | |
| *** ktsuyuzaki has quit IRC | 06:26 | |
| *** zhill has quit IRC | 06:26 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:26 | |
| *** kota_ has quit IRC | 06:28 | |
| *** kota_ has joined #openstack-swift | 06:31 | |
| *** ktsuyuzaki has quit IRC | 06:32 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:35 | |
| *** kota_ has quit IRC | 06:37 | |
| *** kota_ has joined #openstack-swift | 06:40 | |
| *** ktsuyuzaki has quit IRC | 06:42 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:42 | |
| *** kota_ has quit IRC | 06:44 | |
| *** kota_ has joined #openstack-swift | 06:45 | |
| *** silor has joined #openstack-swift | 06:45 | |
| *** ktsuyuzaki has quit IRC | 06:46 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:49 | |
| *** bkopilov has quit IRC | 06:50 | |
| *** kota_ has quit IRC | 06:51 | |
| *** kota_ has joined #openstack-swift | 06:54 | |
| *** ktsuyuzaki has quit IRC | 06:56 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:56 | |
| *** kota_ has quit IRC | 06:58 | |
| *** kota_ has joined #openstack-swift | 07:01 | |
| *** ktsuyuzaki has quit IRC | 07:03 | |
| *** ktsuyuzaki has joined #openstack-swift | 07:04 | |
| *** kota_ has quit IRC | 07:06 | |
| *** kota_ has joined #openstack-swift | 07:07 | |
| *** ktsuyuzaki has quit IRC | 07:08 | |
| *** bkopilov has joined #openstack-swift | 07:45 | |
| *** bkopilov has quit IRC | 07:50 | |
| *** bkopilov has joined #openstack-swift | 07:50 | |
| *** bkopilov has quit IRC | 07:53 | |
| *** bkopilov has joined #openstack-swift | 07:54 | |
| *** nellysmitt has joined #openstack-swift | 08:00 | |
| *** mmcardle has joined #openstack-swift | 08:07 | |
| *** chlong has quit IRC | 08:07 | |
| *** nellysmitt has quit IRC | 08:12 | |
| *** rledisez has joined #openstack-swift | 08:12 | |
| *** nellysmitt has joined #openstack-swift | 08:12 | |
| *** geaaru has joined #openstack-swift | 08:27 | |
| *** bkopilov has quit IRC | 08:31 | |
| *** jordanP has joined #openstack-swift | 08:48 | |
| *** InAnimaTe has quit IRC | 08:57 | |
| *** briancline has quit IRC | 09:00 | |
| *** early has quit IRC | 09:04 | |
| *** briancline has joined #openstack-swift | 09:06 | |
| *** InAnimaTe has joined #openstack-swift | 09:06 | |
| *** early has joined #openstack-swift | 09:06 | |
| *** jistr has joined #openstack-swift | 09:13 | |
| *** haomaiwang has quit IRC | 10:09 | |
| *** abhirc has quit IRC | 10:16 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Fix param description on _make_request https://review.openstack.org/162592 | 10:24 |
| *** kota_ has quit IRC | 10:33 | |
| *** bkopilov has joined #openstack-swift | 10:38 | |
| *** ppai has quit IRC | 11:02 | |
| *** chlong has joined #openstack-swift | 11:03 | |
| *** SkyRocknRoll has quit IRC | 11:03 | |
| openstackgerrit | Alistair Coles proposed openstack/swift-specs: Updates to encryption spec https://review.openstack.org/154318 | 11:09 |
| *** MrHanachoo has joined #openstack-swift | 11:10 | |
| *** MrHanachoo has quit IRC | 11:11 | |
| notmyname | good morning | 11:14 |
| *** ppai has joined #openstack-swift | 11:16 | |
| notmyname | I'd like to draw your attention to http://lists.openstack.org/pipermail/openstack-dev/2015-March/058164.html | 11:20 |
| notmyname | I don't like where that thread is now. the linked bug is https://bugs.launchpad.net/horizon/+bug/1297173 | 11:21 |
| openstack | Launchpad bug 1297173 in OpenStack Dashboard (Horizon) "Cannot view object details with Ceph backend" [Medium,Confirmed] - Assigned to Rohit Karajgi (rohitk) | 11:21 |
| *** ho has quit IRC | 11:24 | |
| portante | notmyname: We talked about removing X-Timestamp from headers a long time ago, right? | 11:42 |
| notmyname | Jay mentions my name in that thread. when he and I talked, I didn't know the context, and said that it's used in container-sync. but I'm not sure how/why horizon is using it (for reads?) | 11:43 |
| *** ppai has quit IRC | 11:43 | |
| notmyname | I dont' think it can be removed because of container sync, right? | 11:44 |
| portante | Right, but it can be removed when the API is not for container sync | 11:45 |
| *** ppai has joined #openstack-swift | 11:45 | |
| portante | additionally, the bug talks about using the X-Timestamp in the response, and not in the API request, so this sounds like the Ceph RadosGW implementation not handling responses the same way | 11:46 |
| portante | I am not sure we document response headers fully, but it was my understanding (now probably way outdated) that X-Timestamp headers in responses are "part" of the API | 11:47 |
| portante | part of the metadata of an object | 11:47 |
| portante | The Ceph implementation uses mtime | 11:48 |
| portante | And finally, Horizon has a UI for viewing individual objects? fascinating | 11:48 |
| notmyname | I think that's all horizon does for swift: shows a dashboard | 11:49 |
| portante | ah, probably container timestamps then | 11:49 |
| notmyname | the first issue is getting horizon to work properly (should it even care about x-timestamp) | 11:49 |
| portante | sounds useful, last modification of a container | 11:50 |
| notmyname | but I don't want a third-party reimplementation of part of swift to start dictating what sort of swift support is in other openstack projects | 11:50 |
| notmyname | ok, I gotta get breakfast before the ops meetup starts | 11:51 |
| notmyname | I'll be back later | 11:51 |
| portante | notmyname: okay, when you get back ... | 11:52 |
| notmyname | yes? | 11:52 |
| portante | The patch referenced in the bug, 82801 (openstack gerrit), appears to be code for getting object details | 11:52 |
| portante | oy | 11:52 |
| portante | not container details | 11:52 |
| portante | and the ceph bug tracked was opened 8 mos ago, updated 27 days ago, assigned to lpabon (http://tracker.ceph.com/issues/8911) | 11:54 |
| portante | and it looks like it was fixed in the firefly release of ceph, awaiting backport to giant | 11:56 |
| *** Anticimex has quit IRC | 11:57 | |
| *** Anticimex has joined #openstack-swift | 11:57 | |
| *** panbalag has joined #openstack-swift | 12:06 | |
| *** yuanz has joined #openstack-swift | 12:25 | |
| *** yuan has quit IRC | 12:25 | |
| acoles | I created a trello board to track crypto work https://trello.com/b/63l5zQhq/swift-encryption | 12:32 |
| acoles | jrichli: ^^ please add yourself as a member and then add stuff | 12:33 |
| acoles | torgomatic: ^^ | 12:33 |
| acoles | notmyname: ^^ can you add link from the trello swift organization? | 12:34 |
| *** annegentle has joined #openstack-swift | 12:36 | |
| *** aix has joined #openstack-swift | 12:43 | |
| *** mmcardle has quit IRC | 12:49 | |
| *** mahatic has joined #openstack-swift | 12:51 | |
| *** Tahmina has joined #openstack-swift | 12:58 | |
| *** chlong has quit IRC | 13:04 | |
| *** mmcardle has joined #openstack-swift | 13:07 | |
| *** ppai has quit IRC | 13:12 | |
| *** chlong has joined #openstack-swift | 13:17 | |
| tdasilva | good morning | 13:20 |
| openstackgerrit | paul luse proposed openstack/swift: Merge master to feature/ec https://review.openstack.org/162630 | 13:23 |
| *** ppai has joined #openstack-swift | 13:24 | |
| *** wshao has joined #openstack-swift | 13:26 | |
| wshao | hi, what is the mailing list for questions? | 13:28 |
| *** wshao has left #openstack-swift | 13:30 | |
| *** ppai has quit IRC | 13:38 | |
| openstackgerrit | Merged openstack/swift: Small optimization to ring builder. https://review.openstack.org/162335 | 13:55 |
| notmyname | tsg: I just heard from mordred (in person) that liberasure code is straightened out | 13:57 |
| notmyname | acoles: cool (trello board) | 13:58 |
| *** reed has joined #openstack-swift | 13:58 | |
| notmyname | acoles: I don't think I can link it, but maybe you can move it into the swift org? | 13:59 |
| notmyname | http://webapps.stackexchange.com/questions/19271/move-trello-boards-between-organizations | 13:59 |
| acoles | notmyname: i like that it has so many fewer features than jira ;) | 13:59 |
| notmyname | :-) | 13:59 |
| notmyname | for all: I propose we skip this week's meeting | 14:00 |
| acoles | notmyname: i already noticed the next agenda is for 18th :) | 14:02 |
| acoles | notmyname: "Note: You'll need to own the board that you're modifying, and you'll need to be a member of the organization that you're transferring the board to" | 14:02 |
| acoles | notmyname: so you need to add me to openstack swift org (i couldn;t see how to self-subscribe) | 14:02 |
| notmyname | acoles: you've now been added to the swift org | 14:02 |
| notmyname | assuming you are acoles on trello | 14:03 |
| acoles | notmyname: dadaa!! (i am) | 14:03 |
| acoles | notmyname: are you further east than usual? | 14:04 |
| notmyname | yes. I'm on the US east coast right now (Pliadelphia) | 14:04 |
| notmyname | Philadelphia | 14:05 |
| notmyname | ops meetup today and tomorrow. then company business on wednesday and home on wed pm | 14:05 |
| acoles | notmyname: ah, not just waking early then. have a good trip. | 14:06 |
| notmyname | ya, I think it will be good | 14:06 |
| notmyname | I like the ops summits since it's a good place to listen to what people are saying | 14:07 |
| *** Anticimex has quit IRC | 14:07 | |
| notmyname | granted, there's not a ton of swift content compared to other ones, but it's good to listen to what people are asking for | 14:07 |
| *** Anticimex has joined #openstack-swift | 14:08 | |
| notmyname | anyone know the updates to http://docs.openstack.org/developer/swift/deployment_guide.html#general-system-tuning | 14:09 |
| notmyname | ahale: ^ | 14:09 |
| *** annegentle has quit IRC | 14:10 | |
| ahale | not off top of my head | 14:10 |
| ahale | there was something recently about some of those needing changing or something wasnt there | 14:10 |
| notmyname | yes | 14:11 |
| notmyname | and I wanted to link to it in reply to http://lists.openstack.org/pipermail/openstack/2015-March/011824.html | 14:11 |
| notmyname | ahale: could you take a few minutes any reply there, please? | 14:11 |
| notmyname | hmm..looks like he's already tried those settings | 14:11 |
| ahale | hm yeah thats a weird error | 14:12 |
| *** annegentle has joined #openstack-swift | 14:15 | |
| openstackgerrit | Merged openstack/swift: Minor cleanup in proxy obj controller https://review.openstack.org/160985 | 14:18 |
| notmyname | these ops summits are fun. sysadmins are not shy about speaking up about what's broken :-) | 14:19 |
| *** nellysmitt has quit IRC | 14:24 | |
| eikke | notmyname: I can confirm :-) | 14:26 |
| *** Anayag has joined #openstack-swift | 14:29 | |
| Anayag | Hi I need some help to understand how PUT request flow. Could you please tell why a PUT request waits at conns = [conn for conn in pile if conn] even the quorum is smaller than the number of connections? | 14:32 |
| notmyname | Anayag: because it's better to cleanly let connections close instead of dropping them on the floor | 14:35 |
| notmyname | Anayag: so once quorum is reached, then there is a secondary (short) delay to allow the other connections to finish, but it will actually drop connections if they don't finish very quickly | 14:36 |
| openstackgerrit | Merged openstack/swift: Ssync fix for EC PUT path https://review.openstack.org/161057 | 14:37 |
| notmyname | lol "anyone have any useful one line debugging tools?" "apt-get update" | 14:39 |
| *** corvus is now known as jeblair | 14:41 | |
| Anayag | +notmyname: Ahh ok, but a scenario I faced was for instance when I have a write quorum of 1, and replication degree is 3. If there is a delay in any one of those 3 replicas it affect for the PUT request its is using a buffered connection? | 14:41 |
| notmyname | if you have 3 replicas, your quorum is 2 | 14:41 |
| Anayag | ya But I changed my the code so it will have read quorum 3 and write quorum 1 to get strong consistency | 14:42 |
| notmyname | that doesn't get you strong consistency | 14:43 |
| Anayag | ahh then how could I achieve strong consistency? Beacause I need to check the system behaviour with strong consistency | 14:44 |
| acoles | peluse: you here? | 14:45 |
| peluse_ | for a bit, yes | 14:45 |
| *** peluse_ is now known as peluse | 14:45 | |
| *** ChanServ sets mode: +v peluse | 14:45 | |
| notmyname | Anayag: use something other than swift if you need strong consistency. | 14:45 |
| notmyname | Anayag: but I'm curious about what you mean by "check the system behaviour..." | 14:46 |
| notmyname | what are you looking for? what are you trying to do? | 14:46 |
| peluse | acoles, what's a hapenin? | 14:46 |
| acoles | peluse: ok just looking at comments on multi frag patch, and concern about if policy ends up being None because policy-index doesn't resolve to a policy... | 14:46 |
| acoles | peluse: seems like an early check raising ann err response would be useful? | 14:46 |
| notmyname | peluse: did you see that the packaging looks like it's taken care of (for liberasurecode)? | 14:47 |
| peluse | acoles, yeah... I'd have to walk back through the scenarios and see if that's an error or not. There are several places where we expect None and assume 0 when that happens but there are a few places where None can never happen. Not sure about this one off the top of my head | 14:47 |
| peluse | notmyname, yeah, just read the scrollback - that's GREAT | 14:47 |
| notmyname | peluse: please check with tsg if there are other outstanding issues. I want to jump on them asap | 14:47 |
| peluse | acoles, for now maybe raising an error is best and we'll find out pretty quick :) | 14:48 |
| acoles | peluse - yep, when index==None we default to 0, thats OK. Its when index = 99, say, and there is no policy 99 | 14:48 |
| Anayag | No I am doing my thesis based on quorum changes and I am checking swift to check at which read/write quorums and which workload scenarios system give best throughput | 14:48 |
| peluse | notmyname, will do | 14:48 |
| tsg | notmyname: ack on liberasurecode .. have been following up with Scott Kitterman (Ubuntu) on that - SpamapS is the one to thank for | 14:48 |
| acoles | because i guess proxy and obj server swift.conf are out of sync?? | 14:48 |
| peluse | ahh, there he is :) | 14:48 |
| Anayag | So my assignment is to find out how swift work for strong consistant situations | 14:49 |
| tsg | good morning! - still waiting on backports for liberasurecode, jerasure and gf-complete for ubuntu 14.04 | 14:49 |
| peluse | acoles, that's one case that got highlighted recently in a few patchs from NT&T, yes. I'm stil a bit ify on coding for miconfigured swift.conf files between servers | 14:49 |
| notmyname | tsg: pypi now has 1.0? | 14:50 |
| notmyname | tsg: and there's no reason to do the frankenstine release where they are all bundled together right? | 14:50 |
| tsg | notmyname: finished that last night - have a call with Kevin at 9am PDT - should be up right after that | 14:50 |
| peluse | tsg, yeah I saw a bunch of commits this weekend, maybe you should take next Sun off (but just one) :) | 14:50 |
| tsg | notmyname: that's correct | 14:50 |
| notmyname | tsg: yay. thanks a ton for that | 14:50 |
| tsg | notmyname: have been following up on zigo and SpamapS on https://bugs.launchpad.net/trusty-backports/+bug/1428834, https://bugs.launchpad.net/trusty-backports/+bug/1428377 | 14:51 |
| openstack | Launchpad bug 1428834 in trusty-backports "[FFE] Please backport jerasure and lib-gf-complete from from Ubuntu 14.10 to 14.04 LTS" [Undecided,New] | 14:51 |
| openstack | Launchpad bug 1428377 in liberasurecode (Ubuntu) "[FFE] Sync new package liberasurecode (for vivid) Backport for Ubuntu 14.04 LTS)" [Wishlist,Triaged] | 14:51 |
| tsg | which made this possible: https://launchpad.net/debian/+source/liberasurecode | 14:51 |
| acoles | peluse: ok so shall i add a check for POLICIES.get_by_index(policy_idx) != None in obj server? | 14:51 |
| tsg | notmyname: and yes, I am thinking the bundled version should be good to have for centos6.5 (not sure when the packages will make their way into that distro) | 14:52 |
| peluse | acoles, specficially where? | 14:52 |
| Anayag | +notmyname: I posted my question in Openstack community web as well https://ask.openstack.org/en/question/62348/why-the-bufferedhttpresponse-stops-proceeding-requests/. I am stuck here not understanding why this happenes? Could you please help? | 14:52 |
| tsg | peluse, notmyname: will post an update to Trello later this morning | 14:52 |
| notmyname | Anayag: it doesn't. swift is AP (in the CAP theorem sense). to do strong consistency (ie CP), there's a lot of changes you'd have to make, including lots of fundamental assumptions in the code | 14:53 |
| mahatic | notmyname, hi! around now. I dozed off the other night (my night). Could you answer that now please? (I'll repost the question) | 14:53 |
| mahatic | notmyname, In validate servers scenario in the recon, the rings would have the info from the config. In which scenario would the ring info be wrong? In other words, why do we need to validate? | 14:54 |
| *** annegentle has quit IRC | 14:54 | |
| acoles | peluse: well, it could go early in each of GET/HEAD/PUT/POST/DELETE or it could go in DiskFile constructor, but then I think we'd need to catch it to return a sanitised error response | 14:54 |
| acoles | peluse: idk! | 14:54 |
| peluse | acoles, maybe we should tackle that in an independent patch? small/compact, single purpose | 14:55 |
| acoles | peluse: sure, i agree, too much changing in one patch otherwise. i'll do that. | 14:56 |
| peluse | acoles, hmmm, get_name_and_placement() will default policy to 0 if None and I think is already in al the entry points | 14:56 |
| notmyname | mahatic: bad info could have been added to the rings, right? that's the whole reason for your project :-) | 14:56 |
| *** glange has joined #openstack-swift | 14:57 | |
| *** ChanServ sets mode: +v glange | 14:57 | |
| mahatic | notmyname, yeah. I'm aware that's the reason :) But I couldn't fit all the things in, now that I read about config info. | 14:58 |
| acoles | peluse: yup that would be a good place to add a check | 14:58 |
| notmyname | mahatic: so I'm not sure I understand the question | 14:58 |
| peluse | acoles, yup, just check against what local POLICIES thinks right there | 15:00 |
| peluse | acoles, versus what came in from the header | 15:00 |
| acoles | peluse: ok will do. | 15:01 |
| notmyname | Anayag: I think that's normal. as in, the proxy will attempt to get a response from each server it's sending data to. and if it gets a quorum of responses, then it will wait just a little bit longer for any other connections to finish | 15:01 |
| mahatic | notmyname, it's more of a config question I believe(the whole paste deploy thing). I'm reading on that. I think I'm fine for now. Will post back if I need to. Sorry to bother :) | 15:01 |
| notmyname | mahatic: no bother at all :-) | 15:01 |
| notmyname | please continue to ask | 15:01 |
| Anayag | +notmyname: Thanks a lot. One last thing Please check whether I have understood correctly, forgetting strong consistency, still lets say we need to do two put requests sequentially. Then 2 of 3 replicas replies and it satisfies the quorum and lets assume 3rd replica is slow (eg: 3 seconds) then at the initiation of the 2nd put request it has to wait until the 3rd replica returns from the previous request. Is this the correct | 15:02 |
| mahatic | :) sure, thanks! | 15:02 |
| notmyname | Anayag: all of the writes are done concurrently (from the proxy to the storage nodes) | 15:02 |
| *** jrichli has joined #openstack-swift | 15:04 | |
| * peluse out for a bit.... | 15:04 | |
| *** Fin1te has joined #openstack-swift | 15:07 | |
| *** nellysmitt has joined #openstack-swift | 15:08 | |
| notmyname | acoles: so i've been thinking about meeting times | 15:08 |
| notmyname | http://www.timeanddate.com/worldclock/meetingtime.html?iso=20150325&p1=224&p2=136&p3=240&p4=248&p5=179 | 15:08 |
| Anayag | +notmyname: Yes, but for 3 replicas, 2 out of the 3 concurrent writes are fine. But if the 3rd write had not yet returned and the next put request is already on the fly. The 2nd put request waits at conns = [conn for conn in pile if conn] line. I need to confirm whether this is the correct behaviour? Or I have understood it wrongly? | 15:09 |
| *** Fin1te has quit IRC | 15:10 | |
| *** Fin1te has joined #openstack-swift | 15:11 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 15:11 |
| *** mahatic has quit IRC | 15:14 | |
| notmyname | Anayag: what do you mean by "next put request"? the next segment of the current session? or just some other request | 15:17 |
| *** annegentle has joined #openstack-swift | 15:23 | |
| Anayag | +notmyname : I am running 2 put requests sequentially. For the first request everything works fine. But when second request start storage node is delayed to return response. | 15:24 |
| notmyname | Anayag: any chance you're using the proxy server's support for ssl? | 15:24 |
| Anayag | no its not using ssl | 15:25 |
| notmyname | ok (good) | 15:25 |
| *** mahatic has joined #openstack-swift | 15:25 | |
| notmyname | err...good that you aren't using the proxy server's ssl | 15:26 |
| acoles | notmyname: reach any conclusion? | 15:28 |
| *** rdaly2 has joined #openstack-swift | 15:28 | |
| notmyname | acoles: "it's hard" | 15:30 |
| notmyname | acoles: actually, though, you can help. what are acceptable times for you? | 15:30 |
| acoles | notmyname: right :) | 15:30 |
| acoles | notmyname: selfish preference would be start time between 8am and 9pm start time. NB cschwede is one hour later than me | 15:33 |
| notmyname | ah, good call. I thought you were both CET | 15:33 |
| acoles | notmyname: No CET is London+1 | 15:34 |
| acoles | notmyname: daylight savings is a pain because I think mattoliverau gets earlier while europe gets later for northern hemisphere summer | 15:35 |
| cschwede | you’re thinking to reschedule meeting time? | 15:35 |
| *** hurricanerix has joined #openstack-swift | 15:36 | |
| notmyname | cschwede: thinking. not deciding yet. acoles expressed displeasure about the current meeting time. and with (the US) time change yesterday, it's come to mind again | 15:36 |
| notmyname | cschwede: so yeah, your input is greatly appreciated here | 15:37 |
| acoles | notmyname: seriously, current time is not a problem for me given other participants, i was most likely being tongue in cheek | 15:37 |
| notmyname | :-) | 15:37 |
| *** david-lyle_afk is now known as david-lyle | 15:38 | |
| notmyname | it's painful for mattoliverau too. and just flat impossible for ho, kota, dmorita, etc | 15:38 |
| notmyname | http://www.timeanddate.com/worldclock/meetingtime.html?iso=20150325&p1=224&p2=136&p3=240&p4=248&p5=179&p6=37 (added Berlin) | 15:38 |
| cschwede | well, unfortunately there is always someone who needs to get up early/stay awake late | 15:39 |
| notmyname | with one meeting time, perhaps 2100UTC is better. late for cschwede and early for japan | 15:39 |
| notmyname | but maybe the least painful overall | 15:39 |
| cschwede | notmyname: that’s fine with me, because then the chance is lower that i’m bringing my kids to bed and i miss the meeting like last time ;) | 15:40 |
| notmyname | cschwede: acoles: hmm...at least until your time change. at which point it gets _really_ late for you | 15:44 |
| notmyname | here's a random day in june | 15:44 |
| notmyname | http://www.timeanddate.com/worldclock/meetingtime.html?month=7&day=25&year=2015&p1=224&p2=136&p3=152&p4=248&p5=179&p6=37&iv=0 | 15:44 |
| acoles | notmyname: thats a very random day in *june* :D | 15:45 |
| notmyname | whatever. I just picked something that starts with a J ;-) | 15:45 |
| Anayag | +notmyname: Could you tell me whether BufferedHttpResponse block a request until a previous connection response return? | 15:46 |
| cschwede | 11pm sounds better than 4am/5am ;) | 15:46 |
| *** nellysmitt has quit IRC | 15:49 | |
| tdasilva | notmyname: what time was last summer? 2000UTC? | 15:49 |
| *** Tahmina has quit IRC | 15:52 | |
| *** nellysmitt has joined #openstack-swift | 15:52 | |
| *** reed has quit IRC | 15:55 | |
| Anayag | +notmyname: Yes It stops until the previous connection response returns | 16:03 |
| *** annegentle has quit IRC | 16:06 | |
| *** tab__ has joined #openstack-swift | 16:07 | |
| *** nellysmitt has quit IRC | 16:11 | |
| openstackgerrit | Merged openstack/swift: Merge master to feature/ec https://review.openstack.org/162630 | 16:15 |
| *** nellysmitt has joined #openstack-swift | 16:21 | |
| *** tab__ has quit IRC | 16:24 | |
| *** wshao has joined #openstack-swift | 16:24 | |
| *** geaaru has quit IRC | 16:27 | |
| *** Trixboxer has quit IRC | 16:29 | |
| *** Trixboxer has joined #openstack-swift | 16:29 | |
| *** wshao has quit IRC | 16:31 | |
| *** wshao has joined #openstack-swift | 16:33 | |
| *** geaaru has joined #openstack-swift | 16:40 | |
| *** zhill has joined #openstack-swift | 16:41 | |
| *** Fin1te has quit IRC | 16:42 | |
| *** reed has joined #openstack-swift | 16:47 | |
| *** dfg has joined #openstack-swift | 16:53 | |
| *** ChanServ sets mode: +v dfg | 16:53 | |
| *** rdaly2 has quit IRC | 16:54 | |
| *** wshao has quit IRC | 16:54 | |
| *** wshao has joined #openstack-swift | 16:55 | |
| *** Krast has quit IRC | 16:55 | |
| *** Krast has joined #openstack-swift | 16:56 | |
| *** Anayag has quit IRC | 16:56 | |
| *** gyee has joined #openstack-swift | 17:02 | |
| *** tsg_ has joined #openstack-swift | 17:04 | |
| *** tsg_ has quit IRC | 17:05 | |
| *** tsg_ has joined #openstack-swift | 17:06 | |
| *** tgohad has joined #openstack-swift | 17:06 | |
| *** tgohad has quit IRC | 17:06 | |
| *** annegentle has joined #openstack-swift | 17:07 | |
| *** tsg has quit IRC | 17:07 | |
| *** nellysmitt has quit IRC | 17:11 | |
| *** nellysmitt has joined #openstack-swift | 17:13 | |
| *** nellysmitt has quit IRC | 17:17 | |
| *** thumpba has joined #openstack-swift | 17:20 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Return 400 from object server if policy does not exist https://review.openstack.org/162696 | 17:24 |
| acoles | peluse: clayg: ^^ trying to catch any possibility of policy == None early and handle gracefully | 17:25 |
| *** rledisez has quit IRC | 17:29 | |
| notmyname | tdasilva: the meeting time has always been 1900utc | 17:36 |
| tdasilva | notmyname: ok | 17:38 |
| *** rdaly2 has joined #openstack-swift | 17:41 | |
| *** wshao has quit IRC | 17:48 | |
| *** reed has quit IRC | 17:49 | |
| notmyname | talking about logs here | 17:51 |
| *** annegentle has quit IRC | 17:55 | |
| *** annegentle has joined #openstack-swift | 17:55 | |
| notmyname | the operator view is that we could take tokens (format strings) for each value and put that in a config | 17:56 |
| notmyname | maybe oslo.log has defined some of these | 17:56 |
| notmyname | and the default (for swift) would be the current log format | 17:57 |
| notmyname | I guess the trick would be to get the necessary tokens defined in oslo.log | 17:57 |
| *** jistr has quit IRC | 17:59 | |
| *** mmcardle has quit IRC | 18:00 | |
| *** chlong has quit IRC | 18:05 | |
| *** mahatic has quit IRC | 18:08 | |
| *** annegentle has quit IRC | 18:08 | |
| *** lcurtis has joined #openstack-swift | 18:15 | |
| *** Fin1te has joined #openstack-swift | 18:25 | |
| *** jordanP has quit IRC | 18:31 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 18:55 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Diskfile decides if durable is written based on policy https://review.openstack.org/162717 | 18:55 |
| *** geaaru has quit IRC | 18:56 | |
| acoles | clayg: peluse : ^^ sorry guys i guess i rebased https://review.openstack.org/159637 on latest feature/ec so created a new patchset :/ | 18:57 |
| *** acoles is now known as acoles_away | 18:59 | |
| clayg | that's fine | 19:00 |
| *** wshao has joined #openstack-swift | 19:13 | |
| *** silor1 has joined #openstack-swift | 19:14 | |
| *** silor has quit IRC | 19:16 | |
| *** bkopilov has quit IRC | 19:17 | |
| *** wshao has quit IRC | 19:29 | |
| *** thumpba has quit IRC | 19:37 | |
| *** thumpba has joined #openstack-swift | 19:38 | |
| *** annegentle has joined #openstack-swift | 19:46 | |
| *** os1 has joined #openstack-swift | 19:51 | |
| os1 | Hi | 19:51 |
| os1 | Has anyone run into | 19:51 |
| os1 | LookupError: No section 'cache' (prefixed by 'filter') found in config /etc/swift/object-expirer.conf | 19:51 |
| os1 | when creating a separate /etc/swift/memcache.conf file? | 19:52 |
| os1 | And the [filter:cache] sections are still there in proxy-server.conf and object-expirer.conf | 19:53 |
| *** bgmccollum has joined #openstack-swift | 19:56 | |
| *** silor1 has quit IRC | 19:58 | |
| bgmccollum | is there any functional difference between presenting swift a whole block device, vs. a single partition on said block device? | 19:58 |
| notmyname | bgmccollum: filesystem partition? | 19:59 |
| bgmccollum | mkfs /dev/device vs. paritioning, then formatting the partition | 19:59 |
| notmyname | os1: I don't know. I'm hoping someone (not at a conference) can spin up a server and test it | 20:00 |
| notmyname | bgmccollum: no semantic difference. swift defines a "node" by IP:port/mountpoint. | 20:01 |
| notmyname | bgmccollum: of course, I don't htink that's a good idea, in general :-) | 20:01 |
| bgmccollum | right...just wondering in the docs, it says to partition the disk, then format the partition, then mount the partition and build ring... | 20:01 |
| bgmccollum | can you expand on why its a bad idea? | 20:02 |
| bgmccollum | or is this fundamentally below swift | 20:02 |
| clayg | bgmccollum: it's just below what swift cares about, you don't need a partition - just a mounted file system | 20:03 |
| clayg | I totally put the fs on the whole device | 20:03 |
| bgmccollum | just wondering if there was something i was missing to why someone would want to create a single partition for the whole device... | 20:03 |
| bgmccollum | i mkfs the whole device myself as well | 20:03 |
| notmyname | bgmccollum: and since, as clayg it's below what swift cares about, you don't want to have 2 partitions as different "devices" where one hardware failure can cause more than one replica loss | 20:04 |
| bgmccollum | notmyname yeah i got that :) which brings up the whole...why even partition in the first place | 20:04 |
| bgmccollum | notmyname: sorry probably misinterpreted your "i dont think that's a good idea"...thought you were saying not a good idea to *not* partition... | 20:06 |
| clayg | bgmccollum: i'm not sure what docs you're looking it - it may be specific to one way of doing a swift-all-in-one where you're just using a single device for four file-systems - or maybe the practice was lifted from there and put somewhere else where it's just not useful | 20:06 |
| bgmccollum | clayg: http://docs.openstack.org/juno/install-guide/install/apt/content/swift-install-storage-node.html - which has been copypasta'd into the RPC docs...im trying to clear up this minconception | 20:08 |
| bgmccollum | "Each of the devices, /dev/sdb and /dev/sdc, must contain a suitable partition table with one partition occupying the entire device." | 20:08 |
| openstackgerrit | Denis Cavalcante proposed openstack/swift: Add storage policy support for sorting method https://review.openstack.org/162743 | 20:09 |
| clayg | yeah that wording is overly dramatic, i'm not sure the history on that line... | 20:09 |
| bgmccollum | so its bunk...thanks for the confirmation | 20:10 |
| clayg | looks new in juno, icehouse just described the process with an example using a partition and made no specific claims as to why -> http://docs.openstack.org/icehouse/install-guide/install/apt/content/installing-and-configuring-storage-nodes.html | 20:11 |
| clayg | well it still said "should be partitioned" - so yeah, urban myth ;) | 20:12 |
| bgmccollum | *massive eye roll* | 20:15 |
| bgmccollum | thanks | 20:15 |
| *** rdaly2 has quit IRC | 20:25 | |
| *** wshao has joined #openstack-swift | 20:30 | |
| *** aix has quit IRC | 20:31 | |
| *** tab has joined #openstack-swift | 20:33 | |
| *** tab is now known as Guest40655 | 20:33 | |
| *** wshao has quit IRC | 20:35 | |
| *** zaitcev has joined #openstack-swift | 20:59 | |
| *** ChanServ sets mode: +v zaitcev | 20:59 | |
| mattoliverau | Morning | 21:05 |
| *** Tahmina has joined #openstack-swift | 21:09 | |
| os1 | notmyname : Are you suggesting that memcached.conf hasn't been tested before? | 21:17 |
| openstackgerrit | paul luse proposed openstack/swift: Erasure Code Reconstructor https://review.openstack.org/131872 | 21:17 |
| *** Fin1te has quit IRC | 21:27 | |
| jrichli | mattoliverau: morning! | 21:28 |
| *** chlong has joined #openstack-swift | 21:45 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: wip: just use policy everywhere https://review.openstack.org/162775 | 21:49 |
| *** chlong has quit IRC | 21:57 | |
| *** chlong has joined #openstack-swift | 21:57 | |
| *** wshao has joined #openstack-swift | 22:01 | |
| *** chlong has quit IRC | 22:04 | |
| *** wshao has quit IRC | 22:06 | |
| *** jrichli has quit IRC | 22:10 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Return 400 from object server if policy does not exist https://review.openstack.org/162696 | 22:16 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: wip: just use policy everywhere https://review.openstack.org/162775 | 22:16 |
| *** wshao has joined #openstack-swift | 22:16 | |
| *** annegentle has quit IRC | 22:19 | |
| *** annegentle has joined #openstack-swift | 22:21 | |
| *** theanalyst has quit IRC | 22:21 | |
| *** openstack has joined #openstack-swift | 22:25 | |
| *** annegentle has quit IRC | 22:26 | |
| *** annegentle has joined #openstack-swift | 22:27 | |
| *** rdaly2 has joined #openstack-swift | 22:29 | |
| *** rdaly2 has quit IRC | 22:33 | |
| peluse | clayg, you there? | 22:45 |
| clayg | yessir | 22:45 |
| peluse | so I'm wasting just tons of time in rebase hell... if I need to rebase *my* patch that is already dependent on someone else's and they update theirs, I just git rebase -i <hash> of their latest right? | 22:46 |
| clayg | so it's "theirs -> yours" and you still want that, but "theirs" has just moved? | 22:47 |
| clayg | the only trick is that when you "git rebase -i new-sha-of-theirs" you have to delete "old-sha-of-theirs" from your "list of shas to rebase ontop of new-sha-of-theirs" | 22:48 |
| clayg | I have pretty good luck with that anwyay... | 22:48 |
| peluse | clayg, ahhh, yes I was forgetting to do that. so the only thing in the pick list should be mine then right? | 22:49 |
| clayg | yup | 22:50 |
| peluse | crap, OK, thanks! | 22:50 |
| *** lcurtis has quit IRC | 22:51 | |
| peluse | good grief.... much better | 22:52 |
| peluse | now I'm wondering if I can add a 2nd dependency that happens to also be a dependency of my parent? Or maybe I should quit while I'm ahead and just cherry pick the code that I want on over... | 22:53 |
| mattoliverau | Lol | 22:53 |
| clayg | so it's "theirs -> yours1 -> yours2" - do the same thing; but start @ yours2 and rebase on "new-sha-of-theirs" and delete "old-shaw-of-theirs" | 22:54 |
| clayg | rebase -i will stop along the way and let you clean up conflicts | 22:54 |
| clayg | the only thing I struggle with going that route is when there's a symatic conflict in yours1 that git does see (like a unused import or failing unittest) | 22:55 |
| clayg | peluse: sometimes it is easier to get yours1 where you're happy - then checkout yours2 and rebase it on yours1 same as you did for "their-new-sha -> yours1" | 22:56 |
| clayg | YMMV | 22:56 |
| clayg | mattoliverau: don't laugh - use DCS they said, it's better they said, makes branching easy they said, phffft | 22:57 |
| peluse | clayg, yeah that might be a better stepwise approach for things that are all entagled | 22:57 |
| mattoliverau | Sorry, I'm just happy to see it isn't just me who gets himself into these messes :p | 22:58 |
| clayg | I think I'm going to just have to make the call to change the kwarg signature on get_diskfile - master passes in policy_idx, but we should pass in the policy | 22:58 |
| clayg | heh | 22:58 |
| peluse | I'll get the recon fully rebased and working tomorrow if it kills me. yeah, that's be one more thing that will throw me a a beer - I mean a loop :) | 22:58 |
| clayg | oh god - no peluse don't *die* | 22:58 |
| clayg | drinking is ok tho | 22:59 |
| peluse | heh... allright time to commute the other way.... later | 22:59 |
| mattoliverau | peluse: o/ | 22:59 |
| *** annegentle has quit IRC | 23:03 | |
| *** os1 has left #openstack-swift | 23:03 | |
| *** Guest40655 has quit IRC | 23:10 | |
| *** km has joined #openstack-swift | 23:21 | |
| *** openstack has joined #openstack-swift | 23:25 | |
| *** chlong has joined #openstack-swift | 23:43 | |
| *** Tahmina has quit IRC | 23:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!