| *** csd has joined #openstack-swift | 00:09 | |
| *** shri has joined #openstack-swift | 00:13 | |
| *** matsuhashi has joined #openstack-swift | 00:26 | |
| *** shri has quit IRC | 00:27 | |
| *** openstackgerrit has joined #openstack-swift | 00:30 | |
| *** mmcardle has joined #openstack-swift | 00:31 | |
| *** mmcardle has quit IRC | 00:35 | |
| *** RockKuo has quit IRC | 00:42 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Merge container storage_policy_index values. https://review.openstack.org/83942 | 00:44 |
|---|---|---|
| *** matsuhashi has quit IRC | 00:47 | |
| *** matsuhashi has joined #openstack-swift | 00:48 | |
| *** csd has quit IRC | 00:51 | |
| *** zackf has quit IRC | 00:51 | |
| *** matsuhashi has quit IRC | 00:52 | |
| *** matsuhashi has joined #openstack-swift | 00:57 | |
| *** mmcardle has joined #openstack-swift | 01:32 | |
| *** zhiyan_ is now known as zhiyan | 01:32 | |
| *** mmcardle has quit IRC | 01:37 | |
| openstackgerrit | A change was merged to openstack/swift: Only run flake8 on Swift code https://review.openstack.org/83177 | 01:46 |
| *** nosnos has joined #openstack-swift | 01:47 | |
| *** changbl has joined #openstack-swift | 01:47 | |
| portante | notmyname: seems pretty clear to me that the testFileSizeLimit functional test does not actually test that file size limits are enforced | 01:54 |
| portante | if you change an SAIO to use 2MB max_file_size in /etc/swift.conf, the test fails | 01:55 |
| portante | sorry, hold on, hopefully it is pilot error | 01:56 |
| portante | well, at least it does not appear to be, changed /etc/swift/test.conf to have max_file_size=2MB and the test does not appear to work, so might be a test harness problem | 01:58 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Merge container storage_policy_index values. https://review.openstack.org/83942 | 02:04 |
| *** RockKuo has joined #openstack-swift | 02:32 | |
| *** mmcardle has joined #openstack-swift | 02:33 | |
| *** mmcardle has quit IRC | 02:38 | |
| *** gyee has quit IRC | 02:39 | |
| openstackgerrit | Yuan Zhou proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 02:44 |
| *** zanc has quit IRC | 02:52 | |
| *** madhuri has joined #openstack-swift | 02:57 | |
| madhuri | clayg: there? | 02:57 |
| clayg | madhuri: it's best to just ask whatever you have on your mind, even if i'm not around right then I'll see it later - unless someone else beats me to it - what's up? | 03:09 |
| *** Anju has joined #openstack-swift | 03:11 | |
| madhuri | clayg: Hi.. | 03:11 |
| clayg | hi :) | 03:11 |
| madhuri | clayg: Just saw your cleanup patch for container-info | 03:12 |
| Anju | clayg : in code in every put of a container the account_update method is calling | 03:13 |
| madhuri | It can be done in account-info case too | 03:13 |
| clayg | yeah - there's been a few other's too - turns out the container-info was big hit! | 03:13 |
| madhuri | for me too :) | 03:13 |
| clayg | madhuri: nice | 03:13 |
| clayg | Anju: ohia | 03:13 |
| madhuri | thanks | 03:13 |
| clayg | madhuri: yeah thank you | 03:13 |
| madhuri | so can I do it for account-info too? | 03:14 |
| clayg | madhuri: yeah i think so | 03:14 |
| madhuri | clayg: ok | 03:14 |
| Anju | clayg : https://github.com/openstack/swift/blob/master/swift/container/server.py#L292 | 03:17 |
| *** vilobhmm has joined #openstack-swift | 03:17 | |
| madhuri | clayg: need some more pointers :) | 03:17 |
| Anju | clayg : according to this in every put of any object and container | 03:18 |
| openstackgerrit | Peter Portante proposed a change to openstack/swift: Update callback with proper bytes transferred https://review.openstack.org/84610 | 03:18 |
| clayg | I feel like I have two outstanding half asked questions... | 03:18 |
| Anju | the account_updater method is changing the timestamp | 03:18 |
| clayg | Anju: so that line you linked is execute by the container server only on container create - nothing to do with objects, or the account updater? | 03:20 |
| Anju | clayg : According to me in every put of any object and any container the account_stat table field "put_timestamp" | 03:21 |
| Anju | should change | 03:21 |
| *** matsuhashi has quit IRC | 03:21 | |
| Anju | clayg : the code line I gave you is saying the same | 03:22 |
| clayg | Anju: no... i don't think so - you have to watchout with some swift clients - they'll do a container PUT before every object PUT | 03:23 |
| clayg | not that that is so bad... | 03:24 |
| Anju | clyag : you mean it is going correct according to you | 03:24 |
| *** zackf has joined #openstack-swift | 03:24 | |
| Anju | clayg: sorry I am taking your timr. I did using API | 03:25 |
| clayg | Anju: so I'm not seeing a container's put_timestamp change when I upload an object - unless I use like the swift cli which does a container PUT before the object PUT - which of course updates the containers put_timestamp as expected? | 03:26 |
| clayg | Anju: i'm not sure I follow your question - are you asking if the account databases should be updated on container create (yes) or if the containers put_timestamp is updated on PUT (yes) or if I can duplicate what you're seeing where object PUT updates container put_timestamp (no) | 03:27 |
| clayg | madhuri: more pointers? | 03:28 |
| Anju | clayg : yeah . yaeh | 03:28 |
| madhuri | clayg: too work upon ;) | 03:28 |
| clayg | oh - lol | 03:28 |
| *** mmcardle has joined #openstack-swift | 03:34 | |
| *** nosnos has quit IRC | 03:37 | |
| clayg | madhuri: swift.common.direct_client is missing a few methods and could use some more/better test coverage? | 03:37 |
| clayg | Anju: so did I accidently answer your questions? | 03:38 |
| madhuri | clayg: Ok. Taking a look | 03:38 |
| Anju | clayg : I want to give you some description . :) | 03:39 |
| *** mmcardle has quit IRC | 03:39 | |
| madhuri | clayg: Yeah. Methods like put_account, head_Account and many more can be addded. | 03:43 |
| madhuri | Right?? | 03:43 |
| clayg | madhuri: yeah! | 03:44 |
| clayg | madhuri: it's just an idea, if you're interested in the backend apis' - and tests | 03:44 |
| *** peluse has joined #openstack-swift | 03:44 | |
| madhuri | clayg: Thanks!! | 03:44 |
| madhuri | Yeah. I can do that | 03:45 |
| Anju | clayg: http://fpaste.org/90826/10383139/ :) | 03:46 |
| peluse | anticw: sorry was offline all day. No, what's there is the latest post but we'll be udpating soon... need to get policies wrapped up | 03:46 |
| peluse | clayg: how are you doing on catching up torgomatic's patch set and looking at those last few outstanding gerrit reviews? Anything I can do to help? | 03:47 |
| clayg | peluse: i was ooo yesterday - i'm currently working on a shared pre-req to both our patches - which is a functioning status_changed_at field so we can say who had the "oldest" storage policy index | 03:49 |
| clayg | peluse: did you ever figure out what was up with those account db tests? sorry i'm behind on ec reviews... | 03:50 |
| peluse | clayg: cool. yes, I posted a fix to that failure. An existing sql trigger was catching our attempt to UPDATE the contaeiner DB | 03:53 |
| peluse | update with SPI of 0 that is, for pre-SPI container tables | 03:53 |
| clayg | peluse: hrmm... i need to look at that change again... | 03:54 |
| clayg | Anju: I'm not seeing it... what do you mean "No change in Acount Database. No change in put_timestamp" | 03:55 |
| Anju | clayg: no chane in account_stat table | 03:55 |
| peluse | gonna go watch a little Game of Thrones... let me know when your pre-req will be ready. In the meantime I'll go take another look at the pluggable DB thingy.. | 03:55 |
| peluse | clayg: here it is: https://review.openstack.org/#/c/83840/ | 03:56 |
| Anju | clayg: what I am saying I pasted it here http://fpaste.org/90830/64113491/ | 04:02 |
| clayg | Anju: right the account's put_timestamp doesn't change on container PUT - just like the container's put_timestamp doesn't change on object PUT | 04:09 |
| Anju | clayg : agree , but why in code It is happening | 04:10 |
| Anju | clayg: https://github.com/openstack/swift/blob/master/swift/container/server.py#L292 | 04:10 |
| Anju | you can see here in every put of object and container | 04:11 |
| Anju | the account_update method is called | 04:11 |
| clayg | yeah that adds a row to the containers table on the account server? | 04:11 |
| clayg | it's the "if container" block of of the account server's PUT method | 04:12 |
| Anju | yes | 04:12 |
| clayg | you can't both agree with me and disagree with me ;) | 04:13 |
| *** lpabon has quit IRC | 04:13 | |
| clayg | yeah account server looks basically the same "if container -> put_container; else -> update_put_timestamp" | 04:14 |
| clayg | ... but I expected that :D | 04:14 |
| Anju | https://github.com/openstack/swift/blob/master/swift/container/server.py#L163 | 04:14 |
| clayg | right that's an account server PUT for /account/container - so it's a different branch in the account server's PUT method | 04:15 |
| clayg | look at your logs, backend account/container server's PUT's take two forms depending on if it's creating the resource or adding an object to the resource | 04:16 |
| clayg | account has PUT /account and PUT /account/container; container has PUT /account/container and PUT /account/container/obj | 04:16 |
| Anju | yeah yeah | 04:16 |
| Anju | correct | 04:16 |
| clayg | heheheheh | 04:17 |
| Anju | clayg: misleading | 04:17 |
| * Anju anju is slapping herself | 04:17 | |
| Anju | :) | 04:18 |
| clayg | and now you know how that works too! | 04:19 |
| Anju | :) | 04:19 |
| *** Manish_ has joined #openstack-swift | 04:24 | |
| Manish_ | hugokuo: Hi..... | 04:27 |
| Manish_ | Do you know , that of "autocreate" value is set to false...then in this case how user can create Account? | 04:29 |
| *** nosnos has joined #openstack-swift | 04:31 | |
| Manish_ | i know that there is one configuration in proxy server.conf.....allow_account_management, but if this one is even set to false, then is it like that no one can create account? | 04:34 |
| clayg | Manish_: if you're doing manual account managment you typically want a set of proxy servers that your user management system is configured to talk to that will accept the PUT /account request (allow_account_management = true) | 04:35 |
| *** mmcardle has joined #openstack-swift | 04:36 | |
| *** matsuhashi has joined #openstack-swift | 04:36 | |
| clayg | the other option is to have allow_account_management true everywhere with some middleware to handle the authorization of reseller admin requests | 04:37 |
| *** mmcardle has quit IRC | 04:40 | |
| * hugokuo look around .... | 04:44 | |
| * clayg backs into the shadows | 04:45 | |
| *** matsuhashi has quit IRC | 04:50 | |
| *** mmcardle has joined #openstack-swift | 04:58 | |
| *** matsuhashi has joined #openstack-swift | 05:00 | |
| *** ppai has joined #openstack-swift | 05:01 | |
| *** mmcardle has quit IRC | 05:01 | |
| *** zackf has quit IRC | 05:14 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Keep status_changed_at up-to-date with status changes. https://review.openstack.org/84621 | 05:14 |
| openstackgerrit | Pete Zaitcev proposed a change to openstack/swift: Relocate DATADIR to backends https://review.openstack.org/84624 | 05:23 |
| *** mmcardle has joined #openstack-swift | 05:28 | |
| *** mmcardle has quit IRC | 05:33 | |
| *** zaitcev has quit IRC | 05:44 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Keep status_changed_at up-to-date with status changes. https://review.openstack.org/84621 | 05:48 |
| *** sushma has joined #openstack-swift | 06:02 | |
| *** sushma has quit IRC | 06:04 | |
| *** vilobhmm has quit IRC | 06:06 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Keep status_changed_at up-to-date with status changes. https://review.openstack.org/84621 | 06:09 |
| *** psharma has joined #openstack-swift | 06:13 | |
| *** sowmya_ has joined #openstack-swift | 06:19 | |
| *** mmcardle has joined #openstack-swift | 06:29 | |
| *** sungju_ has quit IRC | 06:34 | |
| *** mmcardle has quit IRC | 06:34 | |
| *** tanee-away is now known as tanee | 06:35 | |
| *** saju_m has joined #openstack-swift | 06:39 | |
| *** cschwede has left #openstack-swift | 06:41 | |
| *** cheri has joined #openstack-swift | 06:51 | |
| *** sowmya_ has left #openstack-swift | 06:59 | |
| *** mlipchuk has joined #openstack-swift | 07:01 | |
| *** mmcardle has joined #openstack-swift | 07:31 | |
| *** zhiyan is now known as zhiyan_ | 07:35 | |
| *** mmcardle has quit IRC | 07:36 | |
| openstackgerrit | Yuan Zhou proposed a change to openstack/swift: Update container storage_policy_index when auditing https://review.openstack.org/83093 | 07:37 |
| *** saju_m has quit IRC | 07:42 | |
| *** nacim has joined #openstack-swift | 07:51 | |
| *** tanee is now known as tanee-away | 07:51 | |
| *** saju_m has joined #openstack-swift | 07:52 | |
| *** mmcardle has joined #openstack-swift | 08:07 | |
| *** cschwede has joined #openstack-swift | 08:08 | |
| openstackgerrit | Yuan Zhou proposed a change to openstack/python-swiftclient: Allow to specify storage policy when uploading objects https://review.openstack.org/73920 | 08:22 |
| *** mmcardle has quit IRC | 08:23 | |
| *** nshaikh has joined #openstack-swift | 08:27 | |
| *** mmcardle has joined #openstack-swift | 08:27 | |
| *** wayneeseguin has quit IRC | 08:28 | |
| *** wayneeseguin has joined #openstack-swift | 08:28 | |
| *** j_king has quit IRC | 08:28 | |
| *** j_king has joined #openstack-swift | 08:28 | |
| openstackgerrit | Victor Stinner proposed a change to openstack/python-swiftclient: Fix test_multithreading on Python 3 https://review.openstack.org/84104 | 08:31 |
| *** d89 has joined #openstack-swift | 08:39 | |
| *** psharma has quit IRC | 08:44 | |
| *** saju_m has quit IRC | 08:48 | |
| *** tanee-away is now known as tanee | 08:48 | |
| *** tanee is now known as tanee-away | 08:48 | |
| *** tanee-away is now known as tanee | 08:48 | |
| *** tanee is now known as tanee-away | 08:49 | |
| *** tanee-away is now known as tanee | 08:49 | |
| *** saju_m has joined #openstack-swift | 08:49 | |
| *** nshaikh has quit IRC | 08:57 | |
| *** psharma has joined #openstack-swift | 08:58 | |
| *** chandankumar_ has joined #openstack-swift | 08:59 | |
| *** Midnightmyth has joined #openstack-swift | 09:02 | |
| *** psharma has quit IRC | 09:03 | |
| *** Longgeek_ has joined #openstack-swift | 09:04 | |
| *** mmcardle has quit IRC | 09:07 | |
| *** foexle has joined #openstack-swift | 09:07 | |
| *** mmcardle has joined #openstack-swift | 09:08 | |
| *** zanc has joined #openstack-swift | 09:13 | |
| *** acoles is now known as acoles_away | 09:14 | |
| *** acoles_away is now known as acoles | 09:16 | |
| *** Manish_ has quit IRC | 09:17 | |
| *** saju_m has quit IRC | 09:24 | |
| *** saju_m has joined #openstack-swift | 09:25 | |
| openstackgerrit | Madhuri Kumari proposed a change to openstack/swift: Minor cleanup of swift-account-info https://review.openstack.org/84675 | 09:26 |
| *** psharma has joined #openstack-swift | 09:26 | |
| *** saju_m has quit IRC | 09:31 | |
| *** saju_m has joined #openstack-swift | 09:37 | |
| *** jamie_h has joined #openstack-swift | 09:48 | |
| *** matsuhashi has quit IRC | 09:50 | |
| *** ccorrigan has joined #openstack-swift | 09:53 | |
| *** matsuhashi has joined #openstack-swift | 09:57 | |
| *** jamie_h has quit IRC | 09:57 | |
| *** psharma has quit IRC | 10:08 | |
| *** chandankumar_ has quit IRC | 10:11 | |
| *** mmcardle has quit IRC | 10:14 | |
| *** Anju has quit IRC | 10:24 | |
| *** RockKuo_iPad has joined #openstack-swift | 10:29 | |
| *** RockKuo_iPad has quit IRC | 10:30 | |
| *** RockKuo_iPad has joined #openstack-swift | 10:30 | |
| *** mmcardle has joined #openstack-swift | 10:32 | |
| *** RockKuo_iPad has quit IRC | 10:32 | |
| *** RockKuo_iPad has joined #openstack-swift | 10:33 | |
| *** Trixboxer has joined #openstack-swift | 10:41 | |
| *** RockKuo_iPad has quit IRC | 10:42 | |
| *** RockKuo has quit IRC | 10:47 | |
| *** ppai has quit IRC | 10:56 | |
| *** ppai has joined #openstack-swift | 10:58 | |
| *** matsuhashi has quit IRC | 11:00 | |
| *** Anju has joined #openstack-swift | 11:02 | |
| *** nosnos has quit IRC | 11:04 | |
| *** openstackgerrit has quit IRC | 11:08 | |
| *** openstackgerrit has joined #openstack-swift | 11:08 | |
| *** Longgeek_ has quit IRC | 11:10 | |
| *** matsuhashi has joined #openstack-swift | 11:16 | |
| openstackgerrit | Donagh McCabe proposed a change to openstack/swift: Reclaim containers even if account db is reclaimed https://review.openstack.org/84696 | 11:20 |
| *** mmcardle has quit IRC | 11:36 | |
| *** erlon has joined #openstack-swift | 11:51 | |
| *** acoles has quit IRC | 11:56 | |
| *** tdasilva has joined #openstack-swift | 12:03 | |
| *** cheri has quit IRC | 12:04 | |
| *** ppai has quit IRC | 12:05 | |
| *** mkollaro has joined #openstack-swift | 12:11 | |
| *** mmcardle has joined #openstack-swift | 12:13 | |
| *** saju_m has quit IRC | 12:17 | |
| *** cheri has joined #openstack-swift | 12:21 | |
| *** ppai has joined #openstack-swift | 12:22 | |
| openstackgerrit | Peter Portante proposed a change to openstack/swift: Only fetch a timestamp when needed https://review.openstack.org/84710 | 12:36 |
| openstackgerrit | Victor Stinner proposed a change to openstack/python-swiftclient: Python 3: fix tests on HTTP headers https://review.openstack.org/84711 | 12:37 |
| *** chandankumar_ has joined #openstack-swift | 12:38 | |
| *** tdasilva has quit IRC | 12:41 | |
| *** ppai has quit IRC | 12:41 | |
| *** saju_m has joined #openstack-swift | 12:47 | |
| *** chandankumar_ has quit IRC | 12:51 | |
| *** mmcardle has quit IRC | 12:52 | |
| *** cheri has quit IRC | 12:53 | |
| *** chandan_kumar has quit IRC | 12:55 | |
| *** tdasilva has joined #openstack-swift | 12:57 | |
| *** JelleB is now known as a1|away | 12:58 | |
| *** a1|away is now known as JelleB | 12:58 | |
| *** matsuhashi has quit IRC | 13:02 | |
| *** matsuhashi has joined #openstack-swift | 13:02 | |
| *** dmsimard has joined #openstack-swift | 13:06 | |
| *** matsuhashi has quit IRC | 13:07 | |
| *** matsuhashi has joined #openstack-swift | 13:10 | |
| *** Longgeek_ has joined #openstack-swift | 13:15 | |
| *** acoles has joined #openstack-swift | 13:32 | |
| *** ChanServ sets mode: +v acoles | 13:32 | |
| *** RockKuo has joined #openstack-swift | 13:32 | |
| *** saju_m has quit IRC | 13:36 | |
| *** jamesu has joined #openstack-swift | 13:45 | |
| *** chandan_kumar has joined #openstack-swift | 13:49 | |
| *** judd7 has quit IRC | 13:53 | |
| jamesu | Apologies. I'm a bit new to openstack and swift. I'm interested in Hierarchical Storage. I've been reading about erase coding etc. But what is unclear is does it move data from a high performance area (i.e. SSD) to low performance area (i.e large SATA) | 13:58 |
| *** zaitcev has joined #openstack-swift | 13:59 | |
| *** ChanServ sets mode: +v zaitcev | 13:59 | |
| glange | swift clusters can be all spinning disks, a mix of spinning disks and SSD's, or all SSD's (if you have the budget for that) | 14:13 |
| glange | a mix of both types of discs can get you better performance if the account and container servers run on SSDs | 14:13 |
| glange | the erasure encocoding changes won't change where objects are stored just now they are stored | 14:14 |
| glange | I would imagine object GETs and PUTs might be slower with EC because of the overhead | 14:14 |
| glange | jamesu: does that answer your question ? | 14:15 |
| glange | s/just now/just how/ | 14:15 |
| *** judd7 has joined #openstack-swift | 14:16 | |
| jamesu | Ahh. Ok. Yes it does kind of. | 14:16 |
| jamesu | The reason for the question is due to a large media (video) storage capability that we need | 14:17 |
| jamesu | the most recently added is most frequently accessed. Then there is the long-tail historical data | 14:18 |
| jamesu | the latter requiring a lot less performance. | 14:18 |
| glange | the one big advantage of EC is that it reduces storage requirements for an object, right now swift keeps 3 copies of each object, with EC the total storage required per object would be less than 3x | 14:18 |
| glange | jamesu: sounds like you might want to stand up a caching layer in front of swift | 14:19 |
| glange | or use a cdn? | 14:19 |
| jamesu | Yes there is a CDN in front of it. | 14:19 |
| glange | there isn't any feature in swift that makes reads of "hot files" faster | 14:20 |
| peluse | glange: EC can indeed change where objects are stored by using different media for the drives on the EC policy ring | 14:20 |
| jamesu | peluse: can the historical data be migrated to that area at a pre defined time? | 14:20 |
| peluse | to be clear, policies will let you do that, not EC itself :) but EC will be implemnted as a policy | 14:20 |
| glange | peluse: ok, I just meant that with EC the objects would still be stored either on spinning disks or SSD's just like current swift | 14:20 |
| jamesu | or maybe just via an automatic popularity index | 14:20 |
| peluse | jamesu: that will be possible given the base code we're laying down but its not going be a core feature within swift at the moment | 14:21 |
| peluse | glange: gotcha, just wanted to be clear that if someone wanted a policy of all HDDs or all SSDs and EC on one or the other they could do that | 14:21 |
| jamesu | so some external component will have to migrate the object from one area to another? | 14:21 |
| peluse | jamesu: correct | 14:21 |
| peluse | jamesu: imagine a 'default' policy that stores 3x... | 14:22 |
| glange | peluse: ok, I'll have to look closely at storage policies, sounds like an interesting change | 14:22 |
| peluse | then at some later point in time an agent running within the cluster (or outside but that would take more proxy BW) determines its time to migrate stuff to an EC policy, it just 'moves' the data. It can leave behind a manitest type thing so the app doesn't have to know about the location | 14:23 |
| jamesu | Understood. | 14:23 |
| glange | peluse: is it expected that reads on EC objects will be slower? | 14:23 |
| *** mmcardle has joined #openstack-swift | 14:24 | |
| peluse | glange: that would be great! the more eyes, the better. Right now its all on feature/ec but we're in the middle (clayg/torgomatic/me/others) of refactoring into a set of more reasable/logically ordered patches. That's WIP right now off of one of torgomatics repos - you should have that info in your email | 14:24 |
| peluse | glange: yup, most likely | 14:24 |
| peluse | glange: writes as well | 14:25 |
| glange | yeah, but that's the trade off for less disk space used | 14:25 |
| peluse | hard to guage how much though, so many variables... some of the PUT path is done and the GET path won't be far behind so it won't be too long before we can take some real measuremens and see what kind of impact there is | 14:26 |
| peluse | glange: are you coming to the next hackathon? | 14:26 |
| jamesu | peluse: its quite interesting because we are looking to build 3 geographic sites with identical content libraries. | 14:26 |
| glange | peluse: no, I'm sitting that one out, we will be sending out "top men" though :) | 14:27 |
| *** zackf has joined #openstack-swift | 14:27 | |
| peluse | :) OK, well we expect to have the main IO paths deomnstrable by then and generate some really good discussion | 14:28 |
| glange | jamesu: you might want to look at swift's container sync feature | 14:28 |
| glange | jamesu: that lets you have containers across clusters with identical content | 14:29 |
| jamesu | how quickly does the sync happen? | 14:31 |
| jamesu | ahh i see there is an interval parameter. | 14:31 |
| glange | how well it works depends heavily on how much bandwidth you have between clusters | 14:32 |
| glange | but it's a way to automatically keep the containers in sync | 14:32 |
| *** tanee is now known as tanee-away | 14:33 | |
| *** nshaikh has joined #openstack-swift | 14:38 | |
| *** chandankumar_ has joined #openstack-swift | 14:40 | |
| *** matsuhashi has quit IRC | 14:52 | |
| *** Longgeek_ has quit IRC | 15:04 | |
| *** lpabon has joined #openstack-swift | 15:10 | |
| *** cheri has joined #openstack-swift | 15:13 | |
| notmyname | good morning | 15:15 |
| creiht | ugh... I hate gerrit formatting :/ | 15:18 |
| *** mtaylor is now known as mordred | 15:19 | |
| *** mordred has quit IRC | 15:19 | |
| *** mordred has joined #openstack-swift | 15:19 | |
| *** openstackgerrit has quit IRC | 15:21 | |
| *** jamesu has left #openstack-swift | 15:21 | |
| notmyname | creiht: or lack thereof? | 15:21 |
| *** openstackgerrit has joined #openstack-swift | 15:21 | |
| creiht | notmyname: heh | 15:22 |
| creiht | notmyname: I would prefer it didn't try to do anything at all | 15:22 |
| notmyname | creiht: did you see the mailing list question this morning about HEADs being so expensive when doing small-object PUTs? I read it and thought, "If-None-Match sure would be good there" | 15:22 |
| creiht | lol | 15:22 |
| creiht | or just not doing a head at all? :) | 15:23 |
| creiht | I've always disliked that feature of swiftclient | 15:23 |
| notmyname | I see reasons for it. but they all go away with if-non-match support | 15:24 |
| creiht | yeah but I don't think the head should be the default | 15:25 |
| creiht | if I tell you to put a file, it should put the file :) | 15:25 |
| creiht | if I tell you to put a file, only if it isn't already there, then check first | 15:25 |
| notmyname | swift upload --yes --I-mean-it --really --do-it | 15:25 |
| notmyname | --force | 15:25 |
| notmyname | I think that the range requests on SLOs is the last critical thing to get into Icehouse (https://review.openstack.org/#/c/84483/) | 15:27 |
| notmyname | creiht: have you seen anything else critical? | 15:27 |
| creiht | depends on the definition of critical :) | 15:27 |
| creiht | I would really like to get if-none-match in :) | 15:28 |
| notmyname | ya, me too :-) | 15:28 |
| creiht | I'll look at the rest of the list | 15:28 |
| notmyname | just needs one move +2 | 15:28 |
| creiht | yeah | 15:28 |
| notmyname | *more | 15:28 |
| creiht | https://review.openstack.org/#/c/81646/ | 15:28 |
| creiht | for anyone interested :) | 15:28 |
| *** piyush has joined #openstack-swift | 15:29 | |
| creiht | acccount to account copy would be cool, but I think it still needs a little more work | 15:29 |
| notmyname | ok | 15:29 |
| creiht | it would be nice to get the new -info tools cleaned up | 15:30 |
| notmyname | reminder that we have a team meeting today https://wiki.openstack.org/wiki/Meetings/Swift. Not a lot on the agenda. just a couple of status updates on Icehouse and (hopefully from clayg and/or peluse) on storage policies | 15:30 |
| notmyname | creiht: oh yeah. that one really annoys me (the current state, not the outstanding patch) | 15:30 |
| notmyname | and if we get it in, we dont' have to worry as much about migrations because we havent' released with it yet | 15:30 |
| creiht | right | 15:31 |
| creiht | That's all that stick out in my opinion | 15:32 |
| zaitcev | hey, I started breaking out PBEs https://review.openstack.org/84624 | 15:33 |
| peluse | notmyname: hey did you get that other system we shipped you? | 15:34 |
| peluse | zaitcev: cool, I had the next round of review of that on my TODO list for today (and its my b-day so how about that?). SHould I wait until you have it broken out further to avoid confusing myself? | 15:34 |
| zaitcev | peluse: just review them one by one as they come, please | 15:35 |
| notmyname | peluse: heh, I did. :-) | 15:35 |
| peluse | zaitcev: will do! if its not too much trouble please add me as a reviewer as they become ready so I don't forget | 15:35 |
| zaitcev | How many men did it take to carry | 15:35 |
| peluse | notmyname: way cool, wanted to make sure it didn't get lost in the main :) | 15:36 |
| peluse | main=mail | 15:36 |
| notmyname | https://wiki.openstack.org/wiki/Swift/PriorityReviews updated with the 3 patches needed today | 15:37 |
| creiht | notmyname: I didn't +2 the slo as I would like someone besides us to review it :) | 15:38 |
| notmyname | creiht: right, I saw that :-) | 15:38 |
| notmyname | creiht: while I trust you on that one, I understand. | 15:39 |
| creiht | notmyname: what's the likelyhood that Yuan will finish the cleanup? | 15:39 |
| Alex_Gaynor | Can I bug any of ya'll for reviews on https://review.openstack.org/#/c/84205/ and https://review.openstack.org/#/c/84024/ :-) | 15:39 |
| creiht | is he on irc? | 15:39 |
| notmyname | peluse: do you know? | 15:39 |
| creiht | Alex_Gaynor: are those critical for icehouse? :) | 15:39 |
| notmyname | Alex_Gaynor: no, not until the other 3 are done ;-) | 15:39 |
| Alex_Gaynor | creiht: Probably not, not unless Py3k is critical | 15:40 |
| Alex_Gaynor | I hope what they lack in urgency, they make up in simplicity :P | 15:40 |
| creiht | haha | 15:41 |
| Alex_Gaynor | (in all seriousness, if folks are focusedd on more urgent stuff, I totally understand) | 15:41 |
| creiht | Alex_Gaynor: yeah we are trying to focus to get things done to cut a release candidate | 15:42 |
| notmyname | Alex_Gaynor: ya, we need to get these last patches for icehouse in today | 15:42 |
| notmyname | to have an rc tomorow am | 15:42 |
| notmyname | ok, coffee cup empty. time to go to the office | 15:43 |
| notmyname | I'll be back in 30-60 minutes | 15:43 |
| *** tanee-away is now known as tanee | 15:46 | |
| *** Anjut has joined #openstack-swift | 15:47 | |
| creiht | notmyname: I'm tempted to just clean it up for him | 15:48 |
| creiht | but I'm also about to go to lunch, so I'll check in when I get back :) | 15:48 |
| *** mkollaro has quit IRC | 15:49 | |
| creiht | peluse: so you know Yuan Zhou? | 15:50 |
| *** Anjut has quit IRC | 15:51 | |
| peluse | creiht: for sure | 15:53 |
| acoles | creiht: question re https://review.openstack.org/#/c/81646/ ? | 15:54 |
| openstackgerrit | Donagh McCabe proposed a change to openstack/swift: Reclaim containers even if account db is reclaimed https://review.openstack.org/84696 | 15:54 |
| * peluse reads the last bit of history, was off reviewing something | 15:54 | |
| creiht | peluse: re: https://review.openstack.org/#/c/84456/ | 15:55 |
| creiht | do you think they will be able to finish the cleanup? | 15:55 |
| creiht | to get it in today | 15:55 |
| creiht | if not, would they mind a little help? | 15:55 |
| peluse | he's in China so he's probably about to hit the sack right now... I'm certain he wouldn't mind help, he's a really good guy | 15:56 |
| *** mkollaro has joined #openstack-swift | 15:57 | |
| openstackgerrit | Donagh McCabe proposed a change to openstack/swift: Reclaim containers even if account db is reclaimed https://review.openstack.org/84696 | 15:57 |
| *** zhiyan_ is now known as zhiyan | 15:59 | |
| creiht | also, I just noticed that just running `swift setup.py develop` creates a pbr-0.8.0-py2.7.egg file in the swift dir, which is kinda ugly :/ | 15:59 |
| gholt | It's for your own good, I'm sure. | 16:00 |
| creiht | heh | 16:00 |
| peluse | whatever you do don't delete it! :) | 16:00 |
| creiht | too late | 16:00 |
| creiht | :) | 16:00 |
| creiht | I'll see if I can clean it up an push a patch up over his | 16:03 |
| peluse | cool | 16:03 |
| *** mkollaro has quit IRC | 16:04 | |
| creiht | He'll still get the credit ;) | 16:05 |
| *** bill_az has joined #openstack-swift | 16:06 | |
| *** gvernik has joined #openstack-swift | 16:17 | |
| *** chandankumar_ has quit IRC | 16:20 | |
| notmyname | creiht: thanks :-) | 16:22 |
| notmyname | creiht: if you've got that, I'll start on reviewing the SLO range patch | 16:25 |
| *** nshaikh has quit IRC | 16:25 | |
| notmyname | creiht: do you think that having the common code in swift/cli is a good place? | 16:25 |
| openstackgerrit | A change was merged to openstack/swift: Only fetch a timestamp when needed https://review.openstack.org/84710 | 16:29 |
| *** tanee is now known as tanee-away | 16:32 | |
| *** tanee-away is now known as tanee | 16:33 | |
| *** piyush has quit IRC | 16:34 | |
| *** piyush has joined #openstack-swift | 16:37 | |
| *** nacim has quit IRC | 16:38 | |
| *** piyush1 has joined #openstack-swift | 16:39 | |
| *** piyush has quit IRC | 16:41 | |
| *** zhiyan is now known as zhiyan_ | 16:51 | |
| notmyname | cool. so in iTerm2, if you right-click on a number, it shows you the hex eqiv in the context menu | 16:56 |
| *** csd has joined #openstack-swift | 16:58 | |
| *** mlipchuk has quit IRC | 17:13 | |
| notmyname | portante: cschwede: zaitcev: will any of you have a chance to review the if-none-match patch today? https://review.openstack.org/#/c/81646/ | 17:17 |
| notmyname | ah. acoles may be on it :-) | 17:17 |
| acoles | notmyname: i just reviewd it, few minor queries | 17:17 |
| acoles | notmyname: meeting time is 7pm UTC right? | 17:18 |
| notmyname | acoles: 1900utc. | 17:18 |
| notmyname | mathing... | 17:18 |
| notmyname | yes. 7pm utc | 17:18 |
| acoles | notmyname: serious conflict with champions league soccer in europe :) | 17:19 |
| *** mmcardle has quit IRC | 17:21 | |
| notmyname | acoles: I have a follow up to one of your questions there | 17:21 |
| notmyname | acoles: re the spec saying that it should be either * or an etag | 17:21 |
| acoles | notmyname: yeah, i was curious about that | 17:22 |
| *** d89 has quit IRC | 17:23 | |
| notmyname | acoles: I think you're right that it's either * or "1#entity-tag" | 17:23 |
| notmyname | but | 17:23 |
| notmyname | what's the difference? if we accept * as one of many values passed in, a failure means "we have something there already" (ie 412). if we return a 400 instead, it means malformed request | 17:24 |
| notmyname | right? | 17:24 |
| notmyname | there is no functional difference in that the client gets a 4xx error. | 17:25 |
| notmyname | so I don't know how important that difference is (also, the principle of being liberal with what you accept) | 17:26 |
| notmyname | peluse: acoles: also, in response to the question of supporting etag matching on the object server, even if filtered in the proxy, might this be useful later eg for replication or EC rebuilding? | 17:27 |
| notmyname | acoles: so I think you're technically correct (the best kind of correct) wrt tot he spec. good catch! but how important is that in this case? (I don't know) | 17:28 |
| acoles | notmyname: re *, etag - agreed, 4xx goes back either way. It was a question rather than an objection. And I'm not sure if * only is widely enforced elsewhere. | 17:29 |
| *** cheri has quit IRC | 17:29 | |
| *** ankit_deshwal has joined #openstack-swift | 17:30 | |
| *** patchbot` has joined #openstack-swift | 17:31 | |
| *** patchbot has quit IRC | 17:31 | |
| acoles | notmyname: re. etag matching in the obj server, happy for it to stay there. but would be nice to have the extra unit test i suggested. unless i missed something. | 17:32 |
| *** patchbot` is now known as patchbot | 17:33 | |
| *** ankit_deshwal has quit IRC | 17:34 | |
| *** piyush1 has quit IRC | 17:34 | |
| *** tanee is now known as tanee-away | 17:35 | |
| *** hurricanerix_ has joined #openstack-swift | 17:36 | |
| *** nprivalova has quit IRC | 17:38 | |
| *** hurricanerix has quit IRC | 17:38 | |
| acoles | notmyname: creiht: heading out for commute, i'll look at https://review.openstack.org/#/c/81646/ later today | 17:41 |
| openstackgerrit | A change was merged to openstack/swift: AssertEquals is deprecated, use assertEqual partI https://review.openstack.org/76472 | 17:42 |
| zaitcev | muahahaha, I managed to switch Swift meeting reminder to UTC in Zimbra | 17:44 |
| zaitcev | Thanks, Obam... er. VmWare? or Yahoo? | 17:45 |
| *** nprivalova has joined #openstack-swift | 17:47 | |
| *** donagh has joined #openstack-swift | 17:47 | |
| notmyname | acoles: creiht: I think the if-none-match parsing has actually shown a bug in swob (/cc redbo) | 17:48 |
| notmyname | in swob, Match objects have values that _must_ be separated by ', ' (note the space) and cannot distinguish between the special case of a value of * or the entity tag "*" | 17:49 |
| notmyname | which raises the question, "now that it's harder to fix, how much do we care?" | 17:50 |
| *** donagh has quit IRC | 18:02 | |
| openstackgerrit | Madhuri Kumari proposed a change to openstack/swift: Added missing direct backend APIs https://review.openstack.org/84819 | 18:03 |
| *** vilobhmm_ has joined #openstack-swift | 18:07 | |
| *** zul has quit IRC | 18:09 | |
| *** vilobhmm___ has joined #openstack-swift | 18:11 | |
| *** vilobhmm_ has quit IRC | 18:12 | |
| *** zul has joined #openstack-swift | 18:12 | |
| creiht | notmyname: yeah I was pondering the location, and that seemed good enough | 18:13 |
| creiht | just got back from lunch and going to try to finish this up | 18:13 |
| *** vilobhmm___ has quit IRC | 18:17 | |
| cschwede | notmyname: yep, will start reviewing https://review.openstack.org/#/c/81646/ right now and continue after the meeting if required | 18:18 |
| openstackgerrit | Chuck Thier proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 18:36 |
| creiht | notmyname: -^ | 18:36 |
| creiht | there's a quick attempt to clean it up a bit | 18:36 |
| acoles | notmyname: creiht: ah, so '*, etag' is ok but not '*,etag' :) | 18:36 |
| creiht | *, etag doesn't really make sense anyways | 18:37 |
| *** chandankumar_ has joined #openstack-swift | 18:41 | |
| creiht | so I updated some comments on the if-none-match review | 18:43 |
| creiht | if I need to make any changes, I'll fix the typo | 18:44 |
| creiht | :) | 18:44 |
| creiht | unless anyone feels strong enough that it is absolutely required :) | 18:44 |
| *** piyush has joined #openstack-swift | 18:45 | |
| cschwede | notmyname: regarding the differentation between * and "*" - it works with double quotes: from swift.common.swob import Match; print Match('bla, ""*""').tags | 18:46 |
| creiht | well and would "*" ever be a valid etag in swift? | 18:46 |
| *** piyush1 has joined #openstack-swift | 18:47 | |
| cschwede | creiht: don't think so :) | 18:47 |
| creiht | yeah :) | 18:47 |
| openstackgerrit | A change was merged to openstack/swift: Relocate DATADIR to backends https://review.openstack.org/84624 | 18:47 |
| *** piyush has quit IRC | 18:49 | |
| *** shri has joined #openstack-swift | 18:51 | |
| *** chandankumar_ has quit IRC | 18:55 | |
| notmyname | meeting if a few minutes | 18:57 |
| notmyname | creiht: right, not valid from a swift perspective. just a way that swob isn't spec-compliant | 18:58 |
| creiht | yeah | 18:58 |
| creiht | but we can also fix that outside of this patch | 18:58 |
| notmyname | right | 18:58 |
| notmyname | I'll take a look at your comments on the patch after the meeting | 18:59 |
| notmyname | (and potentially lunch) | 18:59 |
| notmyname | swift meeting time | 19:00 |
| cschwede | creiht: I'm wondering why https://review.openstack.org/#/c/81646/6/test/unit/proxy/controllers/test_obj.py doesn't cover lines 667-674 in https://review.openstack.org/#/c/81646/6/swift/proxy/controllers/obj.py | 19:01 |
| creiht | cschwede: hehe yeah | 19:01 |
| creiht | I was just looking into that :) | 19:01 |
| cschwede | creiht: :) | 19:01 |
| zaitcev | creiht: "In combination with "Expect: 100-Continue" this allows the server" -- that is where the switch to Requests turns and bits your behind | 19:04 |
| creiht | zaitcev: heh indeed | 19:04 |
| zaitcev | our collective behind actually | 19:04 |
| *** dmsimard has quit IRC | 19:07 | |
| *** donagh has joined #openstack-swift | 19:08 | |
| creiht | cschwede: heh... so I put a return HTTPServiceUnavailable(request=req) in that block, and the tests fail | 19:09 |
| creiht | so the code is getting executed | 19:10 |
| creiht | zaitcev: luckily for the customer I'm working on this for, they are using .net which does work | 19:10 |
| creiht | 100-continue support is something that we do need to figure out for the requests stuff | 19:11 |
| *** dmsimard has joined #openstack-swift | 19:13 | |
| *** dmsimard1 has joined #openstack-swift | 19:16 | |
| *** dmsimard has quit IRC | 19:17 | |
| *** JuanManuelOlle has joined #openstack-swift | 19:22 | |
| *** fbo is now known as fbo_away | 19:30 | |
| notmyname | creiht: i'll take a look at the patches after I grab some lunch | 19:36 |
| cschwede | creiht: argh, this makes me crazy. if i put a "print" in line 668 in swift/proxy/controllers/obj.py it won't be executed running the unit tests | 19:37 |
| cschwede | creiht: so coverage is right, but then the code is executed when actually used (ie outside unit tests) | 19:37 |
| notmyname | cschwede: isnt' that nose eating stdout? | 19:37 |
| cschwede | notmyname: no, called nose with "nosetest -s" and added also a print right before that to make sure i don't miss anything | 19:38 |
| notmyname | ah | 19:38 |
| *** judd7_ has joined #openstack-swift | 19:38 | |
| *** donagh has quit IRC | 19:38 | |
| *** gvernik has quit IRC | 19:39 | |
| creiht | cschwede: hrm | 19:39 |
| cschwede | creiht: hmm, got it, wait a sec | 19:40 |
| *** judd7 has quit IRC | 19:40 | |
| cschwede | creiht: if i add two brackets around the statement the print command is executed in the unit tests. but coverage still doesn't notice it | 19:43 |
| cschwede | if req.if_none_match is not None and ('*' in req.if_none_match): | 19:43 |
| creiht | cschwede: that doesn't make sense | 19:44 |
| cschwede | creiht: eh, forget what i said. somethings broken here. | 19:44 |
| cschwede | sorry for the noise... | 19:44 |
| creiht | hehe | 19:45 |
| cschwede | creiht: well, anyways. coverage report is somehow wrong, code looks good to me and works as expected, and tests are executed | 19:48 |
| creiht | yeah... I was just double checking the functional tests to make sure they work as expected | 19:48 |
| creiht | it is weird | 19:48 |
| openstackgerrit | Chuck Thier proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 19:52 |
| *** csd has quit IRC | 19:52 | |
| creiht | had to fix a couple of small typos | 19:52 |
| cschwede | creiht: I'll just add my "+2" to the patch without approving - in case someone else wants to have a look before merging | 19:53 |
| creiht | k | 19:53 |
| creiht | thx | 19:53 |
| acoles | cschwede: hah! i had same idea :) | 19:53 |
| creiht | lol | 19:53 |
| acoles | creiht: about my unit test comment... | 19:54 |
| creiht | acoles: yes | 19:54 |
| cschwede | acoles: hehe, so if we put both a 0.5 to the approve it will get merged ;-) | 19:54 |
| acoles | cschwede: or tease creiht with .4999 :) | 19:55 |
| acoles | creiht: ok, so its late here and i may have brain fatigue, but i'm not seeing it | 19:55 |
| creiht | acoles: ok, it is possible I misunderstood your question | 19:55 |
| openstackgerrit | A change was merged to openstack/python-swiftclient: Removed usage of tuple unpacking in parameters. https://review.openstack.org/84024 | 19:55 |
| acoles | i.e. a second put to existing object with if-none-match that does not match so therefore resulting in a 201 response | 19:56 |
| creiht | oh | 19:56 |
| creiht | with '*' it will never put if there is any object there | 19:56 |
| acoles | creiht: yes. i'm referring to the non-* case at the object server (since obj server is matching against etags) | 19:57 |
| creiht | oh | 19:57 |
| creiht | heh | 19:57 |
| creiht | yeah forgot this was the object server | 19:57 |
| *** tsg has joined #openstack-swift | 19:58 | |
| acoles | creiht: :) | 19:58 |
| acoles | anyways...i decided on my commute that since its not functionality that is exposed (yet) that the test could wait, for the sake of progressing this | 19:58 |
| creiht | acoles: thx | 19:59 |
| acoles | creiht: cschwede: so one remaining nit is notmyname's observation about whitespace in the etag list (even though etag lists have no value here :) | 20:00 |
| creiht | lol seems my hasty refactor doesn't past flake8 | 20:00 |
| cschwede | creiht: do you mind if i update the the clean up patch? Some unused imports in request_helpers | 20:00 |
| creiht | cschwede: by all means | 20:00 |
| creiht | acoles: yeah, but that should be fixed outside that patch | 20:00 |
| creiht | as it is a bug in swob | 20:00 |
| acoles | creiht: cool | 20:01 |
| cschwede | acoles: yep, i had some troubles with common.swob.Match as well | 20:01 |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 20:04 |
| creiht | cschwede: thx | 20:05 |
| cschwede | creiht: you're welcome. but i was to hasty, adding an empty line in request_helpers. oh my... | 20:06 |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 20:07 |
| creiht | lol | 20:08 |
| openstackgerrit | A change was merged to openstack/swift: Range requests not working with sub_SLOs https://review.openstack.org/84483 | 20:08 |
| *** mkollaro has joined #openstack-swift | 20:08 | |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 20:08 |
| cschwede | ok, it's definitive too late here. I'l continue tomorrow morning... | 20:09 |
| creiht | cschwede: hah... thanks for the help today | 20:09 |
| cschwede | creiht: no problem, you're welcome! let me know if there is an urgent review i can do in around 10 hours | 20:10 |
| creiht | will do | 20:11 |
| creiht | but it is looking like things are winding down | 20:11 |
| cschwede | nice :) see you all, have a nice day! bye | 20:13 |
| *** fifieldt has quit IRC | 20:20 | |
| *** fifieldt_ has joined #openstack-swift | 20:20 | |
| *** RockKuo_TW has joined #openstack-swift | 20:35 | |
| *** JuanManuelOlle has quit IRC | 20:35 | |
| *** RockKuo has quit IRC | 20:38 | |
| notmyname | if-none-match patch marked as approved | 20:39 |
| notmyname | creiht: are you looking at or pushing anything to the swift*info patch? | 20:40 |
| *** JuanManuelOlle has joined #openstack-swift | 20:40 | |
| creiht | notmyname: well I'm a bit conflicted since I contributed to it :) | 20:42 |
| creiht | and my allergies are killing me | 20:42 |
| notmyname | yikes. oak is killer today http://saallergy.info | 20:42 |
| creiht | yup | 20:42 |
| notmyname | creiht: just checking if it's ready for me to review or if you have something outstanding | 20:42 |
| creiht | and we made the mistake of eating outside at chuy's for lunch today | 20:42 |
| creiht | ahh | 20:42 |
| creiht | notmyname: as far as I know it should be ready | 20:43 |
| notmyname | (not that all of your work isn't outstanding) | 20:43 |
| notmyname | ok | 20:43 |
| creiht | lol | 20:43 |
| tdasilva | notmyname: just heard about the Swift June Hackathon, is that open to non-core developers? | 20:43 |
| notmyname | tdasilva: it is! https://www.eventbrite.com/e/openstack-swift-june-hackathon-tickets-8309569145 | 20:44 |
| tdasilva | nice...i'd like to attend...we'll see | 20:45 |
| tdasilva | thx | 20:45 |
| *** rmitchell has joined #openstack-swift | 20:47 | |
| rmitchell | I'm trying to troubleshoot a 503 error from my swift cluster | 20:47 |
| *** JuanManuelOlle has quit IRC | 20:47 | |
| rmitchell | I've looked in /var/log/messages | 20:47 |
| rmitchell | nada | 20:47 |
| rmitchell | I'm using Keystone for Auth | 20:48 |
| rmitchell | I can do every action but when I attempt to delete a container I get not found | 20:49 |
| rmitchell | Can anybody give me some pointers ? | 20:49 |
| rmitchell | Swift is hard ;) | 20:50 |
| *** a_hacker has joined #openstack-swift | 20:50 | |
| *** csd has joined #openstack-swift | 20:50 | |
| notmyname | rmitchell: have you reconfigured syslog to send the messages somewhere else | 20:53 |
| notmyname | ? | 20:53 |
| notmyname | rmitchell: where are you getting the 503? what operation? | 20:53 |
| *** tdasilva has left #openstack-swift | 20:55 | |
| *** lpabon has quit IRC | 20:56 | |
| *** JuanManuelOlle has joined #openstack-swift | 20:59 | |
| *** JuanManuelOlle1 has joined #openstack-swift | 21:03 | |
| rmitchell | <+notmyname> I've added the following to the proxy-server.conf | 21:03 |
| rmitchell | # Logging set log_name = proxy-server set log_facility = LOG_LOCAL0 set log_level = DEBUG set log_address = /var/log/proxy-server.log | 21:04 |
| *** JuanManuelOlle has quit IRC | 21:04 | |
| rmitchell | And run swift-init proxy restart | 21:04 |
| notmyname | rmitchell: I don't think that will work for log address | 21:07 |
| rmitchell | makes sense - I'm not seeing a log file | 21:07 |
| rmitchell | ;) | 21:07 |
| *** JuanManuelOlle has joined #openstack-swift | 21:08 | |
| notmyname | rmitchell: ya, that's for the syslog device (or you can set it up for udp logging to a remote host) | 21:08 |
| notmyname | rmitchell: you'll need to configure rsyslogd.conf or syslogng.conf (or whatever you're using) to redirect the logs out to a custom file | 21:09 |
| rmitchell | ok - and /dev/log just goes to syslog | 21:09 |
| *** JuanManuelOlle1 has quit IRC | 21:09 | |
| notmyname | right | 21:10 |
| notmyname | rmitchell: eg http://docs.openstack.org/developer/swift/development_saio.html#optional-setting-up-rsyslog-for-individual-logging | 21:10 |
| rmitchell | I've gone back to /dev/log. Assuming the default will put into syslog | 21:10 |
| *** bvandenh has joined #openstack-swift | 21:12 | |
| *** Trixboxer has quit IRC | 21:15 | |
| notmyname | creiht: can you share any more about the .net user who needs the if-none-match support? what's their use case? | 21:17 |
| *** bvandenh has quit IRC | 21:20 | |
| notmyname | Info is uncomplete! | 21:22 |
| *** JuanManuelOlle has quit IRC | 21:23 | |
| *** dmsimard has joined #openstack-swift | 21:23 | |
| *** dmsimard1 has quit IRC | 21:24 | |
| *** sungju_ has joined #openstack-swift | 21:36 | |
| *** foexle has quit IRC | 21:37 | |
| rmitchell | <+notmyname> got past the 503. Had a misconfigured proxy (which has now been shutdown). I can create a container, but swift reports back a 404. However I can then turn around and put an object in the container and I get a 404 | 21:39 |
| notmyname | rmitchell: so you PUT to a container, get a 404, and then try to PUT and object in the container and get a 404? | 21:39 |
| notmyname | you'll get a 404 on a container PUT if the account doesn't exist. the rest sounds normal to me | 21:40 |
| rmitchell | When I shell, I see part of the URI on disk | 21:40 |
| rmitchell | I think the account must exist - I'm using it to connect and interact with the cluster | 21:41 |
| rmitchell | We're utilmately trying to get to the point of running cosbench - it terminates with 404 | 21:42 |
| notmyname | rmitchell: your keystone account isn't the same thing as a swift account. pay no attention to the common substring of "account" ;-) | 21:45 |
| rmitchell | I'm obviously confused ;) We are creating accounts on Keystone. We are configuring Swift to use Keystone for authentication. | 21:47 |
| rmitchell | We are able to auth with keystone and then swift stat | 21:48 |
| notmyname | and what do you see when you do that? | 21:48 |
| notmyname | what's the value of account_autocreate in your proxy config files? | 21:48 |
| rmitchell | [root@sprswiftprox01 envs]# swift stat Account: AUTH_838dec975ad948e1aab29c203f5dfb48 Containers: 7 Objects: 0 | 21:49 |
| notmyname | ok, that's good | 21:49 |
| notmyname | so `swift post container` doesn't work? | 21:49 |
| rmitchell | account_autocreate = true | 21:49 |
| rmitchell | [root@sprswiftprox01 swift]# swift post container Container PUT failed: http://sprswiftprox01.stni.seagate.com:8080/v1/AUTH_838dec975ad948e1aab29c203f5dfb48/container 404 Not Found [first 60 chars of response] <html><h1>Not Found</h1><p>The resource could not be found.< | 21:50 |
| rmitchell | [root@sprswiftprox01 swift]# swift stat Account: AUTH_838dec975ad948e1aab29c203f5dfb48 Containers: 8 Objects: 0 Bytes: 0 Accept-Ranges: bytes X-Timestamp: 1396290904.79357 Content-Type: text/plain; charset=utf-8 | 21:50 |
| rmitchell | [root@sprswiftprox01 swift]# swift list container containerz containter2 cosbench_trouble mycontainers_1 new_test testing testing2 | 21:51 |
| *** sungju_ has quit IRC | 21:51 | |
| rmitchell | container put failed, but swift stat then lists it ? | 21:51 |
| *** sungju_ has joined #openstack-swift | 21:52 | |
| rmitchell | another example | 21:52 |
| rmitchell | [root@sprswiftprox01 swift]# swift post notmyname Container PUT failed: http://sprswiftprox01.stni.seagate.com:8080/v1/AUTH_838dec975ad948e1aab29c203f5dfb48/notmyname 404 Not Found [first 60 chars of response] <html><h1>Not Found</h1><p>The resource could not be found.< [root@sprswiftprox01 swift]# swift stat Account: AUTH_838dec975ad948e1aab29c203f5dfb48 Containers: 9 Objects: 0 Bytes: 0 Accept-Ranges: bytes X-Tim | 21:52 |
| notmyname | paste.openstack.org might be good here ;-) | 21:52 |
| notmyname | rmitchell: ok, let's walk through some API requests and see where it fails | 21:53 |
| rmitchell | gotcha | 21:53 |
| notmyname | rmitchell: you got the logging set up? ie you can see logs now? | 21:53 |
| clayg | paste.openstack.org is never *good* - it's tolerably slow sometimes... | 21:53 |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 21:54 |
| notmyname | clayg: well, that's true | 21:54 |
| notmyname | creiht: added my own patches to the swift-{account,container}-info patch | 21:54 |
| notmyname | creiht: mostly to fix the pep8 and typo in tests bug. but then also so that swift-object-info uses the same ring locations code too | 21:55 |
| *** sungju_ has quit IRC | 21:56 | |
| rmitchell | +notmyname - are you interested in the proxy server logs ? | 21:57 |
| notmyname | rmitchell: mostly just that you can see them :-) | 21:57 |
| rmitchell | yep - seeing handoff requests on the proxy server logs | 21:57 |
| portante | clayg, notmyname: got a fix for constraints comin' | 21:58 |
| portante | in order to change a constraint for testing, unit or functional, each test seems to pick the location that it gets used | 21:58 |
| rmitchell | and 404 errors from the container-server on the storage nodes | 21:58 |
| portante | but this means relies on the binding at module import time | 21:59 |
| notmyname | rmitchell: I think you need to figure out why the handoff nodes are being used | 21:59 |
| notmyname | rmitchell: you've got all the processes running in the cluster, right? ie servers + replication + auditing + etc | 22:00 |
| *** erlon has quit IRC | 22:02 | |
| rmitchell | We've got Nagios monitoring servers and validating the expected numbers of processes | 22:03 |
| *** Midnightmyth has quit IRC | 22:04 | |
| openstackgerrit | Peter Portante proposed a change to openstack/swift: Rework use of straights to ease testing https://review.openstack.org/84877 | 22:05 |
| rmitchell | Does the log entry "proxy-server Handoff requested (1) (txn: tx61ec43fe14ff4f10a524c-00533c84f5) " indicate that a Handoff is actually being used or does it indicate that a handoff is being identified ? | 22:05 |
| openstackgerrit | Peter Portante proposed a change to openstack/swift: Rework use of constraints to ease testing https://review.openstack.org/84877 | 22:06 |
| *** dmsimard has quit IRC | 22:06 | |
| notmyname | rmitchell: being used | 22:06 |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 22:08 |
| notmyname | rmitchell: ok, got that patch taken care of. it was distracting me | 22:09 |
| notmyname | rmitchell: you're getting a 404 in response to the PUT. but it's showing up in the listing. which is strange to me | 22:09 |
| notmyname | rmitchell: I'd suggest walking through some commands with curl so you know exactly what's going on | 22:10 |
| notmyname | rmitchell: start with `swift stat -v` first (assuming you have all the auth creds set in env vars. that will show you the storage url and the auth token you'll need | 22:10 |
| notmyname | rmitchell: eg on my SAIO https://gist.github.com/notmyname/187831cb63bd1078eece | 22:11 |
| notmyname | rmitchell: so from that, I can do this: https://gist.github.com/notmyname/185edb46b169c5fe4345 | 22:12 |
| notmyname | rmitchell: that's a listing followed by a PUT to create a container, then another listing | 22:12 |
| rmitchell | ok - ran swift stat -v | 22:13 |
| notmyname | rmitchell: notice that in each of those responses there is the X-Trans-Id header. that's returned on every response from swift and it uniquely identifies that request/response. so you can grep for it in your logs and see everything that happened | 22:13 |
| rmitchell | ok | 22:14 |
| rmitchell | see that | 22:14 |
| rmitchell | have seen the TX in syslog also | 22:14 |
| creiht | notmyname: who is going to review it if we all make contributions? :) | 22:15 |
| notmyname | rmitchell: so what will be interesting to see is the logs from your container PUT and then from the subsequent container listing | 22:16 |
| notmyname | creiht: :-) | 22:16 |
| notmyname | creiht: I would have gone ahead and put my +2, but I didn't since I also changed a few things (and added the swift-object-info stuff) | 22:16 |
| notmyname | creiht: well it's not like this is patching the ring builder or anything. if you add your +2 I'll add mine and merge it :-) | 22:18 |
| creiht | haha | 22:18 |
| creiht | I was just looking at it | 22:18 |
| *** Jesse22 has joined #openstack-swift | 22:18 | |
| rmitchell | +notmyname Re: http://paste.openstack.org/show/74911/ | 22:18 |
| *** piyush has joined #openstack-swift | 22:18 | |
| *** piyush has left #openstack-swift | 22:19 | |
| *** Jesse22 has quit IRC | 22:19 | |
| rmitchell | When I check on disk I can see the 59824 directory structure on disk on the primary nodes but nothing inside them | 22:19 |
| rmitchell | I do not see anything on the handoff nodes | 22:19 |
| *** piyush1 has quit IRC | 22:20 | |
| notmyname | rmitchell: one of the next things to do is make sure you have the same rings everywhere. eg if you have swift-recon installed (and why wouldn't you?), run `swift-recon --md5` | 22:21 |
| rmitchell | 8/8 hosts matched, 0 error[s] while checking hosts. | 22:23 |
| notmyname | ok, that's good. | 22:23 |
| creiht | notmyname: the ring locations for swift-object-info is wrong | 22:23 |
| notmyname | ah. oops | 22:23 |
| creiht | it is showing it as hash.data instead of ts.data | 22:24 |
| notmyname | bah | 22:24 |
| creiht | haha | 22:24 |
| notmyname | well that makes it not nearly as much fun | 22:28 |
| *** MarkWStorer has joined #openstack-swift | 22:28 | |
| creiht | yeah, I thought of messing with the object stuff but then realized it might be a little trickier than I wanted to mess with at the time :) | 22:28 |
| *** MarkWStorer has quit IRC | 22:28 | |
| rmitchell | <+notmyname> any other suggestions ? | 22:29 |
| notmyname | creiht: ya. I'm reverting that and I'll push it again | 22:29 |
| notmyname | rmitchell: can you paste the logs for a container PUT and then the account listing? | 22:30 |
| *** MarkWStorer has joined #openstack-swift | 22:32 | |
| *** MarkWStorer has quit IRC | 22:32 | |
| notmyname | creiht: ...which actually allows for a little more cleanup | 22:35 |
| creiht | lol | 22:36 |
| notmyname | this is the most complicated simple patch in a while ;-) | 22:38 |
| creiht | hah | 22:38 |
| *** mwstorer has joined #openstack-swift | 22:39 | |
| rmitchell | +notmyname thx - http://paste.openstack.org/show/qZyw42N8PEzkR7F9qQF3/ | 22:42 |
| *** zackf has quit IRC | 22:43 | |
| *** mkollaro has quit IRC | 22:43 | |
| notmyname | rmitchell: looks like the long keystone token is hiding some info. add "reveal_sensitive_prefix = 12" to the proxy config in the proxy-logging section. that way it will only show the first 12 bytes of the auth token | 22:44 |
| notmyname | rmitchell: also, seeing the storage logs too would be helpful | 22:44 |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 22:46 |
| notmyname | creiht: ^^ | 22:46 |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 22:48 |
| notmyname | what's one more patch set? | 22:48 |
| notmyname | I had a typo in the swift-object-info, so I reverted that whole file back to master | 22:49 |
| creiht | lol | 22:49 |
| notmyname | ok, I think I'm good now :-) | 22:50 |
| notmyname | creiht: compare with patch set 7 to see what I actually ended up changing | 22:50 |
| notmyname | hmm...just noticed the bare except with no # noqa | 22:51 |
| creiht | hah | 22:51 |
| rmitchell | +notmyname http://paste.openstack.org/show/AxrUJPmFSq0euWCmVklq/ | 22:52 |
| notmyname | creiht: did you see what rmitchell is doing? POST/PUT to create a container, gets a 404, but its in the listing | 22:53 |
| creiht | notmyname: sorry haven't followed closely | 22:54 |
| notmyname | creiht: well, that's the tldr ;-) | 22:54 |
| creiht | hah | 22:54 |
| notmyname | rmitchell: can you grep your storage node logs for transaction id "tx26b8fa1b186b413b9ac44-00533c9454" ? | 22:54 |
| rmitchell | +notmyname http://paste.openstack.org/show/2kSvHMeIiwuWi8nZ6EVV/ output from storage node | 22:54 |
| notmyname | rmitchell: that's the container create | 22:54 |
| notmyname | ah, ok | 22:54 |
| creiht | notmyname: I'm actually just trying to hold on long enough to get this last patch in | 22:56 |
| creiht | allergies kinda have me down | 22:56 |
| notmyname | creiht: no worries :-) | 22:56 |
| notmyname | creiht: so the reason for the bare except is to test the call to ring. but the test code only passes in [] | 22:57 |
| notmyname | so it gets AttributeError and then the info code raises ValueError | 22:57 |
| creiht | can you just do | 22:58 |
| creiht | except Exception: | 22:58 |
| notmyname | ya, I was thinking except (ValueError, AttributeError): | 22:59 |
| rmitchell | <+notmyname> any ideas ? | 23:01 |
| notmyname | rmitchell: just a second. I'm trying to finish up this last patch for Icehouse with creiht too :-) | 23:02 |
| rmitchell | ok, sorry (blush) | 23:02 |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: Clean up swift-{account, container}-info https://review.openstack.org/84456 | 23:03 |
| notmyname | creiht: ok, there. | 23:03 |
| notmyname | rmitchell: no worries | 23:03 |
| notmyname | rmitchell: maybe I foudn something | 23:05 |
| notmyname | rmitchell: lot's of assumptions here, but maybe it's your issue | 23:06 |
| rmitchell | excited | 23:06 |
| a_hacker | ;) | 23:06 |
| rmitchell | sounds probable | 23:06 |
| rmitchell | <waits with baited breath> | 23:06 |
| notmyname | need to check on emore thing | 23:07 |
| clayg | *bated | 23:07 |
| notmyname | well, maybe not | 23:08 |
| rmitchell | I've been eating fish ;) | 23:08 |
| notmyname | rmitchell: here's what I was thinking....you have a port number wrong in your ring | 23:08 |
| creiht | notmyname: ok looks good to me | 23:09 |
| creiht | I'm out, see everyone tomorrow | 23:09 |
| notmyname | creiht: feel better :-) | 23:09 |
| creiht | heh | 23:09 |
| creiht | hope so | 23:09 |
| notmyname | rmitchell: but looking again, that may not be it. I was seeing the referer log field not the actual request | 23:09 |
| notmyname | rmitchell: ie I was thinking the account server was responding to a container request | 23:10 |
| notmyname | rmitchell: something to check, for sure. but I'm not sure it's the issue now | 23:10 |
| rmitchell | ports look correct on output from lsof on a sample storage node - 6000, 6001, 6002 | 23:11 |
| rmitchell | I'm going to have to bail and digg into this in the morning | 23:15 |
| notmyname | rmitchell: ah ok. a_hacker are you on this too? | 23:16 |
| a_hacker | Yes sir | 23:16 |
| a_hacker | :) | 23:16 |
| rmitchell | thanks for your help in looking into this with us | 23:16 |
| notmyname | sure | 23:16 |
| notmyname | a_hacker: you want to keep looking? | 23:16 |
| a_hacker | I'm about to bail as well | 23:16 |
| notmyname | ah ok | 23:16 |
| rmitchell | Any reccomended reading on tracing a transaction through the system as we have been doing | 23:16 |
| a_hacker | But rest assured you will see us tomorrow or even later tonight | 23:16 |
| notmyname | rmitchell: I think you have it. use the transaction id | 23:17 |
| zaitcev | I saw those phantom 404s when updaters failed to keep up, but in such cases re-posting that container allowed to delete or use it normally. | 23:17 |
| notmyname | rmitchell: a_hacker: seems that the next steps are to simplify the system until it works as expected and then see what goes wrong | 23:17 |
| zaitcev | Fixing updaters first of course. | 23:17 |
| notmyname | a_hacker: rmitchell: eg turn off services and using curl instead of the CLI | 23:18 |
| rmitchell | makes sense | 23:18 |
| rmitchell | thanks again | 23:18 |
| a_hacker | Sounds good, we'll play with it some more. Thanks for your help. | 23:18 |
| notmyname | of course | 23:18 |
| zaitcev | If rmitchell screwed up port assignments, it would be a different error, not 404. I forgot what but it's some kind of Invalid Argument. 420 or something. | 23:18 |
| * notmyname starts to work on CHANGELOG updates for Icehouse then | 23:19 | |
| notmyname | zaitcev: ya, that makes sense | 23:19 |
| notmyname | zaitcev: you thinking of switching over to working on keystone? ;-) | 23:19 |
| notmyname | zaitcev: re your twitter comments :-) | 23:20 |
| openstackgerrit | Jenkins proposed a change to openstack/python-swiftclient: Updated from global requirements https://review.openstack.org/83493 | 23:20 |
| zaitcev | notmyname: I'm surprised because I thought RAX had their own auth. | 23:20 |
| notmyname | zaitcev: ya, but they also invented keystone | 23:20 |
| zaitcev | ah | 23:20 |
| notmyname | RAX is an enigma ;-) | 23:20 |
| openstackgerrit | Jenkins proposed a change to openstack/swift: Updated from global requirements https://review.openstack.org/81207 | 23:20 |
| zaitcev | And gosh guys, I went to look over If-None-Match and it's already merged! | 23:21 |
| zaitcev | Agile devel! | 23:21 |
| notmyname | ok, why do the global requirements need to be bumped up so often? keystone bumps a version that some library wants and we all need a new version of it? what if the old version still works? | 23:21 |
| *** a_hacker has quit IRC | 23:22 | |
| *** rmitchell has quit IRC | 23:22 | |
| notmyname | clearly this release-critical bug of a minor version bump in keystoneclient must be merged immediately! | 23:23 |
| * notmyname can only hold back the snark for so long | 23:23 | |
| zaitcev | you are sarcastic, aren't you | 23:23 |
| notmyname | at least my -2 stays on the swift one. updating keystoneclient in test-requirements on swiftclient. sure, I guess. clearly there are valid and critical reasons this is needed | 23:24 |
| notmyname | let's just go ahead and bump it to >=1.0 so we can just skip over the next several releases | 23:25 |
| *** bada has quit IRC | 23:26 | |
| notmyname | is it appropriate to put a doge meme on a gerrit review? | 23:26 |
| notmyname | too late | 23:26 |
| Alex_Gaynor | I lol'd | 23:27 |
| notmyname | :-) | 23:28 |
| Alex_Gaynor | I think the answer to all questions about syncing wrt to global-requirements is that ~~distros~~ want to make sure they can always install everything in one giant global namespace | 23:28 |
| notmyname | right. I totally get that. I'm not opposed to the idea of global requirements. I'm not a big fan of openstack's current implementation of the idea | 23:30 |
| notmyname | eg if we started using some new feature in eventlet 2.0 (because we wrote the feature and it was brackwards-breaking), does that mean that every openstack project now needs to use eventlet 2.0? (substitute whatever library you want) | 23:31 |
| notmyname | the simpler case (no version brackets) would be massively improved by having a simple version solver | 23:32 |
| notmyname | eg swift's requirements are >= older versions than in global requirements. still works, no conflicts | 23:32 |
| notmyname | (but I will admit we're a little bit lucky in that sense, that some other openstack project or upstream dependency hasn't put an upper bound on some dependency) | 23:33 |
| *** dmsimard has joined #openstack-swift | 23:47 | |
| *** keving has joined #openstack-swift | 23:52 | |
| portante | notmyname: still around? | 23:59 |
| notmyname | portante: yup. what's up? | 23:59 |
| portante | anything left to review? | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!