| *** tsg has quit IRC | 00:14 | |
| *** tsg has joined #openstack-swift | 00:14 | |
| *** ChanServ sets mode: +v jrichli | 00:30 | |
| openstackgerrit | Merged openstack/swift: Updating Bandit config file https://review.openstack.org/365065 | 00:33 |
|---|---|---|
| *** nadeem has quit IRC | 00:36 | |
| *** vint_bra has joined #openstack-swift | 00:41 | |
| *** vint_bra has joined #openstack-swift | 00:50 | |
| *** nadeem has joined #openstack-swift | 01:10 | |
| *** nadeem has quit IRC | 01:15 | |
| *** sgundur_ has joined #openstack-swift | 01:34 | |
| *** sgundur_ has quit IRC | 01:38 | |
| *** zhengyin has joined #openstack-swift | 01:44 | |
| zhengyin | good moring | 01:44 |
| zhengyin | morning | 01:45 |
| charz | morning | 01:56 |
| *** fguillot is now known as fguillot_afk | 02:24 | |
| *** fguillot_afk has quit IRC | 02:26 | |
| *** frekoya_ has joined #openstack-swift | 02:27 | |
| zhengyin | charz:morning | 02:28 |
| *** frekoya_ has quit IRC | 02:30 | |
| mattoliverau | zhengyin, charz: morning | 02:37 |
| *** mingyu has quit IRC | 02:38 | |
| *** janonymous has joined #openstack-swift | 02:57 | |
| *** vint_bra has quit IRC | 02:57 | |
| zhengyin | mattoliverau:morning | 02:58 |
| *** SkyRocknRoll has joined #openstack-swift | 03:21 | |
| *** SkyRocknRoll has quit IRC | 03:26 | |
| *** SkyRocknRoll has joined #openstack-swift | 03:43 | |
| *** mingyu has joined #openstack-swift | 03:53 | |
| *** klrmn has quit IRC | 03:54 | |
| *** mingyu has quit IRC | 03:57 | |
| MooingLemur | "mattoliverau:morning" looks like a user:group specification you'd pass to chown | 04:13 |
| *** psachin_ has joined #openstack-swift | 04:32 | |
| *** ppai has joined #openstack-swift | 04:37 | |
| *** tsg has quit IRC | 04:59 | |
| *** pcaruana has quit IRC | 05:05 | |
| *** ChanServ sets mode: +v cschwede | 05:26 | |
| *** mingyu has joined #openstack-swift | 05:39 | |
| *** ChubYann has quit IRC | 05:40 | |
| *** mingyu has quit IRC | 05:43 | |
| timburke | cnf: it's auth-system dependent. assuming keystone, there's an ec2-credential that can be created that will provide your access/secret keys | 05:50 |
| timburke | try `openstack ec2 credentials create --help` | 05:50 |
| mattoliverau | MooingLemur: isn't it nice that zhengyin decided to use my irc nick as his Linux username and must also be a morning person to have a groupname of morning :P | 05:51 |
| mattoliverau | timburke: your not asleep or weekending very well :P | 05:51 |
| *** rcernin has joined #openstack-swift | 06:03 | |
| *** pcaruana has joined #openstack-swift | 06:17 | |
| *** hseipp has joined #openstack-swift | 06:42 | |
| *** rledisez has joined #openstack-swift | 07:17 | |
| *** oshritf has joined #openstack-swift | 07:18 | |
| *** oshritf_ has joined #openstack-swift | 07:18 | |
| janonymous | notmyname: Hi, i guess you were right when you said we can rethink about adding bandit to gate. Now i realize that.If you say i could remove it or if we choose to go with it then also it is great, but only thing that bothers me is it's maintainance.. | 07:29 |
| *** geaaru has joined #openstack-swift | 07:32 | |
| *** sams-gleb has joined #openstack-swift | 07:49 | |
| *** sams-gleb has quit IRC | 07:49 | |
| *** amitkqed has quit IRC | 07:53 | |
| *** amitkqed has joined #openstack-swift | 07:54 | |
| *** tsg has joined #openstack-swift | 07:55 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Add process level concurrency to container sync https://review.openstack.org/210099 | 07:59 |
| mattoliverau | ^^ Just a rebase, probe tests had diverged a little since it was last updated so had to do a bit on manual conflict merging. But probe tests now pass, so I think is all good :) Now to continue reviewing it :) | 08:00 |
| *** kong has quit IRC | 08:01 | |
| acoles | mattoliverau: thanks for rebasing that ^^ I had just noticed the -1 from Jenkins | 08:05 |
| *** joeljwright has joined #openstack-swift | 08:06 | |
| *** ChanServ sets mode: +v joeljwright | 08:06 | |
| *** treyd has quit IRC | 08:12 | |
| *** tanee has quit IRC | 08:13 | |
| *** hugokuo has quit IRC | 08:13 | |
| *** kencjohnston has quit IRC | 08:13 | |
| *** clarkb has quit IRC | 08:13 | |
| *** portante has quit IRC | 08:13 | |
| *** treyd has joined #openstack-swift | 08:14 | |
| *** hugokuo has joined #openstack-swift | 08:14 | |
| *** sams-gleb has joined #openstack-swift | 08:14 | |
| *** tanee has joined #openstack-swift | 08:14 | |
| *** portante has joined #openstack-swift | 08:14 | |
| *** kencjohnston has joined #openstack-swift | 08:14 | |
| *** clarkb has joined #openstack-swift | 08:15 | |
| *** sgundur1 has joined #openstack-swift | 08:17 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: api-ref: Move repeated paragraph to an include file https://review.openstack.org/367471 | 08:21 |
| acoles | mattoliverau: torgomatic: you had +2's on that ^^ but it fell into merge conflict, now fixed. | 08:23 |
| cnf | timburke thanks! | 08:29 |
| cnf | (no idea how long ago you responded ^^; | 08:29 |
| openstackgerrit | Alistair Coles proposed openstack/swift: api-ref: fix some header definitions https://review.openstack.org/366232 | 08:29 |
| acoles | cschwede: I will merge your patch 366569 into this ^^ | 08:33 |
| patchbot | https://review.openstack.org/#/c/366569/ - swift - api-ref: fix more header definitions | 08:33 |
| *** asettle has joined #openstack-swift | 08:34 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: api-ref: fix some header definitions https://review.openstack.org/366232 | 08:42 |
| acoles | cschwede: ^^ rebased and with your change squashed in | 08:44 |
| acoles | donagh: you had a +2 on previous version of that ^^ | 08:44 |
| openstackgerrit | Alistair Coles proposed openstack/swift: api-ref: clean up account listing examples https://review.openstack.org/366211 | 08:46 |
| *** silor has joined #openstack-swift | 09:26 | |
| *** silor1 has joined #openstack-swift | 09:28 | |
| *** silor has quit IRC | 09:30 | |
| *** silor1 is now known as silor | 09:30 | |
| *** swifterdarrell has quit IRC | 09:42 | |
| *** swifterdarrell has joined #openstack-swift | 09:44 | |
| *** ChanServ sets mode: +v swifterdarrell | 09:44 | |
| *** kong has joined #openstack-swift | 09:49 | |
| *** a1|away has quit IRC | 09:55 | |
| *** pdardeau has quit IRC | 09:56 | |
| *** pdardeau has joined #openstack-swift | 09:56 | |
| *** hoonetorg has quit IRC | 09:56 | |
| *** a1|away has joined #openstack-swift | 09:56 | |
| *** gopenshaw has quit IRC | 09:57 | |
| *** hoonetorg has joined #openstack-swift | 09:57 | |
| *** gopenshaw has joined #openstack-swift | 09:58 | |
| *** mvk has quit IRC | 10:02 | |
| *** openstackstatus has quit IRC | 10:28 | |
| *** mvk has joined #openstack-swift | 10:39 | |
| *** yarkot1 has quit IRC | 11:00 | |
| *** mmotiani has quit IRC | 11:00 | |
| *** mmotiani_ is now known as mmotiani | 11:00 | |
| *** mmotiani_ has joined #openstack-swift | 11:01 | |
| *** mmotiani_ has quit IRC | 11:02 | |
| *** mmotiani_ has joined #openstack-swift | 11:03 | |
| *** mmotiani has quit IRC | 11:04 | |
| *** ntata has quit IRC | 11:04 | |
| *** smurke has quit IRC | 11:04 | |
| *** sgundur1 has quit IRC | 11:04 | |
| *** yarkot has quit IRC | 11:04 | |
| *** mmotiani_ is now known as mmotiani | 11:04 | |
| openstackgerrit | Merged openstack/swift: api-ref: Move repeated paragraph to an include file https://review.openstack.org/367471 | 11:04 |
| *** mmotiani_ has joined #openstack-swift | 11:09 | |
| *** zul has joined #openstack-swift | 11:13 | |
| *** ntata has joined #openstack-swift | 11:14 | |
| *** yarkot has joined #openstack-swift | 11:20 | |
| *** yarkot1 has joined #openstack-swift | 11:20 | |
| *** hoonetorg has quit IRC | 11:23 | |
| *** tsg has quit IRC | 11:25 | |
| *** manous has joined #openstack-swift | 11:31 | |
| *** manous has quit IRC | 11:40 | |
| *** sams-gleb has quit IRC | 11:42 | |
| *** sams-gleb has joined #openstack-swift | 11:42 | |
| *** hoonetorg has joined #openstack-swift | 11:56 | |
| *** hoonetorg has quit IRC | 11:57 | |
| *** hoonetorg has joined #openstack-swift | 11:58 | |
| *** oshritf has quit IRC | 11:58 | |
| *** oshritf_ has quit IRC | 11:58 | |
| *** oshritf_ has joined #openstack-swift | 11:58 | |
| *** oshritf has joined #openstack-swift | 11:58 | |
| *** porunov has joined #openstack-swift | 12:03 | |
| porunov | Hello. How to configure proxy servers to use identity v3? Right now I am using tempauth for authentication. I use identity v1 (curl -v -H 'X-Auth-User: admin:admin' -H 'X-Auth-Key: admin' http://192.168.0.58:8080/auth/v1.0/). Is it possible to configure servers to use tempauth with v3 or v2 identity? This doesn't work: curl -H "Content-Type: application/json" -X POST -d '{"auth" : {"passwordCredentials" : {"username" : "admin","password" : "admin"}," | 12:04 |
| porunov | tenantName" : "admin"}}' http://192.168.0.58:8080/auth/v2.0/ | 12:04 |
| *** manous has joined #openstack-swift | 12:18 | |
| *** vint_bra has joined #openstack-swift | 12:22 | |
| *** tsg has joined #openstack-swift | 12:28 | |
| *** fguillot has joined #openstack-swift | 12:34 | |
| *** david-lyle has joined #openstack-swift | 12:38 | |
| *** david-lyle has quit IRC | 12:43 | |
| *** tsg has quit IRC | 12:45 | |
| *** fguillot is now known as fguillot_afk | 12:47 | |
| *** fguillot_afk has quit IRC | 12:48 | |
| *** fguillot has joined #openstack-swift | 12:52 | |
| *** tsg has joined #openstack-swift | 12:54 | |
| acoles | porunov: docs to use keystone with swift are here might you be able to spare me some time to chat about your experience/tips for working remotely for barefoot? I am nowhere near any decision but thinking I should get myself informed. | 13:02 |
| *** manous has quit IRC | 13:02 | |
| acoles | porunov: http://docs.openstack.org/developer/swift/overview_auth.html#keystone-auth | 13:02 |
| acoles | tempauth does not use v2 or v3 identity API | 13:02 |
| *** tsg has quit IRC | 13:05 | |
| acoles | porunov: with keystone you need to get a token from keystone service by passing credentials and then pass the token to swift, you do not send your id credentials to swift. | 13:06 |
| acoles | porunov: the swiftclient cli does this under the hood for v2 and v3 api (i.e. makes a request to keytone for token, then to swift). If you use swift CLI with the --debug option you will see the equivalent curl commands in the debug output | 13:07 |
| *** asettle has quit IRC | 13:08 | |
| *** asettle has joined #openstack-swift | 13:09 | |
| *** manous has joined #openstack-swift | 13:10 | |
| *** fguillot is now known as fguillot_afk | 13:17 | |
| *** vint_bra has quit IRC | 13:23 | |
| *** vint_bra has joined #openstack-swift | 13:27 | |
| openstackgerrit | Ondřej Nový proposed openstack/swift: Fixed cut and pasted paragraph from commit message in manpage https://review.openstack.org/371932 | 13:34 |
| *** kong has quit IRC | 13:41 | |
| onovy | acoles: thanks | 13:43 |
| acoles | onovy: np | 13:44 |
| *** tsg has joined #openstack-swift | 13:46 | |
| porunov | acoles: Could you explain me how how all requests are managed by proxy when we use keystone? I am not sure that I understand how keystone works. After client have received a token from keystone he have to put it in each request to work with openstack. But as I read about keystone it saves all tokens in a local database. So, if I want to put an object into swift then a client sends request to a proxy server to store an object and the proxy server | 13:47 |
| porunov | sends a request to the keystone server to check a token (each time for each request?). As I understand the performance can suffer if we have some latency between proxy and keystone servers. Am I wrong? I realy don't understand how all requests are managed by proxy servers when we use keystone. Sincerely | 13:47 |
| *** vint_bra has quit IRC | 13:48 | |
| *** tsg_ has joined #openstack-swift | 13:50 | |
| *** vint_bra has joined #openstack-swift | 13:50 | |
| onovy | i think tokens are cached in memcached on proxy servers | 13:50 |
| onovy | in swauth they are :) | 13:50 |
| *** tsg has quit IRC | 13:50 | |
| acoles | porunov: swift uses two pieces of middleware to support keystone - the authtoken middleware (which is part of the keystoneauth project so not in the swift repo) and kesytoneauth middleware (which is specific to swift). authtoken receives the token in client request and validates it with the keystone service, and also caches the token validation result for 5 mins (default time) http://docs.openstack.org/developer/keystonem | 13:52 |
| acoles | iddleware/middlewarearchitecture.html | 13:52 |
| acoles | http://docs.openstack.org/developer/keystonemiddleware/middlewarearchitecture.html | 13:52 |
| acoles | porunov: for a validated token, authtoken populates the request env with identity info (like user name, project etc) which keystoneauth then uses to authorise the request (or not) | 13:53 |
| porunov | acoles: Thank you very much. It is really great. Do you know how to change the time for cache from 5 mins to 60 mins? | 14:00 |
| *** vint_bra has quit IRC | 14:01 | |
| porunov | acoles: I found, thanks. token_cache_time ) | 14:02 |
| acoles | porunov: according to this doc http://docs.openstack.org/developer/keystonemiddleware/middlewarearchitecture.html set the option 'token_cache_time = 3600' in the filter:authtoken section of proxy-server.conf | 14:02 |
| acoles | porunov: yep, that :) | 14:02 |
| *** vint_bra has joined #openstack-swift | 14:04 | |
| acoles | porunov: btw I noticed there is a bug in the doc I linked to you for swift config with keystone https://bugs.launchpad.net/swift/+bug/1604674 | 14:05 |
| openstack | Launchpad bug 1604674 in OpenStack Object Storage (swift) "Doc error in Auth Overview for specifying keystone domain " [Undecided,New] | 14:05 |
| *** oshritf has quit IRC | 14:06 | |
| *** oshritf_ has quit IRC | 14:06 | |
| acoles | porunov: if you follow the openstack install guide, and create a keystone domain with name 'default' for your services, then the correct config for authtoken is as shown here: http://docs.openstack.org/mitaka/install-guide-ubuntu/swift-controller-install.html | 14:07 |
| acoles | porunov: historically the domain had id=default but now it is name=default.should patch the swift docs. | 14:07 |
| * acoles should patch the swift docs | 14:08 | |
| *** tsg_ has quit IRC | 14:15 | |
| *** asettle has quit IRC | 14:15 | |
| *** asettle has joined #openstack-swift | 14:16 | |
| *** tongli has joined #openstack-swift | 14:16 | |
| *** sgundur has joined #openstack-swift | 14:19 | |
| *** sgundur_ has joined #openstack-swift | 14:19 | |
| *** tsg_ has joined #openstack-swift | 14:21 | |
| *** manous has quit IRC | 14:22 | |
| porunov | acoles: Thank you very much for help =) | 14:31 |
| *** sams-gleb has quit IRC | 14:33 | |
| *** manous has joined #openstack-swift | 14:36 | |
| *** openstackgerrit has quit IRC | 14:37 | |
| *** openstackgerrit has joined #openstack-swift | 14:37 | |
| *** SkyRocknRoll has quit IRC | 14:38 | |
| *** david-lyle has joined #openstack-swift | 14:40 | |
| *** david-lyle has quit IRC | 14:45 | |
| *** tsg_ has quit IRC | 14:50 | |
| *** manous has quit IRC | 14:51 | |
| *** manous has joined #openstack-swift | 14:55 | |
| *** juzuluag has joined #openstack-swift | 14:56 | |
| *** manous has quit IRC | 15:00 | |
| *** manous has joined #openstack-swift | 15:01 | |
| *** manous has quit IRC | 15:05 | |
| *** pcaruana has quit IRC | 15:07 | |
| *** ppai has quit IRC | 15:13 | |
| *** psachin_ has quit IRC | 15:18 | |
| *** manous has joined #openstack-swift | 15:18 | |
| *** oshritf has joined #openstack-swift | 15:18 | |
| *** oshritf_ has joined #openstack-swift | 15:19 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Update authtoken domain options https://review.openstack.org/372573 | 15:20 |
| *** manous has quit IRC | 15:23 | |
| *** klrmn has joined #openstack-swift | 15:25 | |
| *** klrmn has quit IRC | 15:27 | |
| *** sgundur_ has quit IRC | 15:27 | |
| *** rcernin has quit IRC | 15:30 | |
| acoles | notmyname: might be worth trying to include that fix ^^ in the next release | 15:30 |
| *** david-lyle has joined #openstack-swift | 15:32 | |
| *** manous has joined #openstack-swift | 15:36 | |
| *** manous has quit IRC | 15:41 | |
| *** Renich has joined #openstack-swift | 15:42 | |
| Renich | hey guys | 15:43 |
| *** manous has joined #openstack-swift | 15:43 | |
| Renich | I am having these issues: https://paste.fedoraproject.org/431042/42998001/ | 15:43 |
| Renich | Any idea what could be the cause? | 15:43 |
| Renich | It is a 2-node Keystone + Swift installation | 15:44 |
| Renich | The storage uses ZFS | 15:44 |
| Renich | just 1 replica | 15:44 |
| *** Jeffrey4l_ has left #openstack-swift | 15:46 | |
| *** sgundur_ has joined #openstack-swift | 15:46 | |
| *** sgundur_ has quit IRC | 15:49 | |
| *** manous has quit IRC | 15:55 | |
| *** sgundur_ has joined #openstack-swift | 15:55 | |
| *** oshritf has quit IRC | 15:56 | |
| *** oshritf_ has quit IRC | 15:56 | |
| *** tsg_ has joined #openstack-swift | 15:57 | |
| openstackgerrit | Mahati Chamarthy proposed openstack/swift: Delete old tombstones https://review.openstack.org/346865 | 15:57 |
| *** zaitcev has joined #openstack-swift | 15:59 | |
| *** ChanServ sets mode: +v zaitcev | 15:59 | |
| *** porunov has quit IRC | 16:01 | |
| *** arch-nemesis has joined #openstack-swift | 16:06 | |
| *** manous has joined #openstack-swift | 16:07 | |
| *** dmorita has joined #openstack-swift | 16:30 | |
| *** sgundur_ has quit IRC | 16:30 | |
| *** dmorita has quit IRC | 16:32 | |
| *** dmorita has joined #openstack-swift | 16:32 | |
| timburke | good morning | 16:33 |
| Renich | o/ | 16:34 |
| openstackgerrit | Merged openstack/swift: Turn on F812 check https://review.openstack.org/371799 | 16:35 |
| *** sgundur_ has joined #openstack-swift | 16:40 | |
| timburke | Renich: is your object-server at 10.0.64.191:6000/1 running? it looks like you're getting timeout from the proxy-server trying to connect to it | 16:40 |
| timburke | also, are you running with only one replica? that will drastically reduce your durability guarantees | 16:41 |
| Renich | timburke: I think it is. The service is up. You mean swift-object.service, right? | 16:41 |
| *** rledisez has quit IRC | 16:41 | |
| *** nadeem has joined #openstack-swift | 16:42 | |
| Renich | timburke: I've had the server lock up. I had to delete all the .lock files and restart the services to make it work again | 16:43 |
| *** markstur_ is now known as markstur | 16:44 | |
| Renich | timburke: my boss is betting on ZFS's deduplication, compression and RAIDZ capabilities | 16:45 |
| Renich | timburke: the use case involves a single dude uploading weather data from a satelite, and multimple readers | 16:46 |
| timburke | yep, swift-object.service sounds right. from the proxy server, if you try `curl http://10.0.64.191:6000/` does the server respond? | 16:46 |
| Renich | timburke: I've warned them about the RAID warning. They say it's irrelevant in our case | 16:46 |
| Renich | timburke: Invalid path: / | 16:47 |
| Renich | it does | 16:47 |
| timburke | hurrah! | 16:47 |
| Renich | ;=) | 16:47 |
| timburke | then why did it time out... :-/ | 16:47 |
| zaitcev | because it took too much time to complete the operation :-) | 16:47 |
| Renich | timburke: I think it's related to ZFS in some way. What if I have low IOPs or something? | 16:47 |
| Renich | zaitcev: heh... yeah | 16:48 |
| Renich | The dude is putting up to 1.4M files in a container | 16:48 |
| Renich | could that have something to do? | 16:48 |
| zaitcev | That, too, isn't going to help matters when proxy is doing get_info. | 16:48 |
| *** ChubYann has joined #openstack-swift | 16:49 | |
| zaitcev | The good news is, all timeouts are configurable. | 16:49 |
| Renich | zaitcev: yeah, I've been mangling with them | 16:49 |
| Renich | up to 30s in some cases | 16:49 |
| zaitcev | I'd cry and headdesk if I were a few years younger but by the age of 50 I learned not to give shit about mistakes of the architects and people like Renich have to make right. | 16:50 |
| zaitcev | s/and people/that people/ | 16:50 |
| *** rcernin has joined #openstack-swift | 16:51 | |
| zaitcev | I once read memoir of a famous tank designers, who created T-72B amoing other things. He described how he's been tasked with the start-up warmer for T-55. It was kinda this situation... The chief designed just handwaved it and said we'll have someone design a warmer, no problem. | 16:52 |
| *** Suyi_ has joined #openstack-swift | 16:53 | |
| zaitcev | I think the problem we're hearing about is very similar to what Gluster people did, e.g. R=1 etc. | 16:53 |
| *** mvk has quit IRC | 16:55 | |
| zaitcev | You still have to put big containers on SSD though, I think. Otherwise timeouts alone aren't fixing it. | 16:55 |
| *** mmotiani has quit IRC | 16:57 | |
| *** yarkot1 has quit IRC | 16:57 | |
| *** mmotiani_ is now known as mmotiani | 16:57 | |
| *** hseipp has quit IRC | 16:58 | |
| MooingLemur | why is handoffs_first undesirable in normal cluster operation? | 16:58 |
| *** mmotiani_ has joined #openstack-swift | 16:59 | |
| openstackgerrit | Merged openstack/swift: remove comment saying we ignore H233. we actually check it https://review.openstack.org/371790 | 16:59 |
| openstackgerrit | Merged openstack/swift: DRY out probe.common https://review.openstack.org/370321 | 17:00 |
| *** mmotiani_ has quit IRC | 17:00 | |
| Renich | zaitcev: LOL, I hear ya! Well, ZFS has some cache drives. 8 SSDs so it captures writes fast enough. Given the info is being sent throught the Internet, these should, at least, help a bit on getting them fast into the cache, then, putting them, slowly, into the spinning drives | 17:00 |
| Renich | we're talking about ~900 TB here | 17:01 |
| Renich | on each server | 17:01 |
| zaitcev | Okay, just make sure you aren't flushing container data with object data. | 17:02 |
| Renich | zaitcev: what do you mean? Like mixing the concepts or putting everything on magnetics? | 17:03 |
| zaitcev | Container needs to be on SSD so that SQLite can look up and update large indexes in a reasonable amount of time. If you keep files that back container DBs and your object files in the same cache, objects can easily crowd container out of the "cache". | 17:04 |
| zaitcev | (subject to many assumptions about how all this works) | 17:05 |
| *** klrmn has joined #openstack-swift | 17:06 | |
| *** oshritf_ has joined #openstack-swift | 17:07 | |
| Renich | zaitcev: OK, so metadata on the DBs have to live there. I might be able to do something to keep them in the cache | 17:07 |
| *** oshritf has joined #openstack-swift | 17:07 | |
| Renich | s@have@has@ | 17:07 |
| *** zul has quit IRC | 17:07 | |
| Renich | zaitcev: the "containers" directory contains those, right? | 17:08 |
| zaitcev | Renich: yes. The file(1) should be able to confirm, I'd expect. | 17:09 |
| *** oshritf_ has quit IRC | 17:09 | |
| *** oshritf has quit IRC | 17:09 | |
| Renich | zaitcev: This is awesome, thanks a lot man. I think this will solve our issues | 17:10 |
| *** geaaru has quit IRC | 17:11 | |
| *** asettle has quit IRC | 17:16 | |
| *** asettle has joined #openstack-swift | 17:22 | |
| MooingLemur | Renich: I tried ZFS as storage, even with zil and l2arc, the iops were substantially amplified compared to something like ext4 with dm-cache, and the ZFS storage nodes were constantly at much higher iowaits than the ext4 ones. | 17:24 |
| zaitcev | [root@rhev-a24c-02 ~]# file /srv/node/b3/containers/197670/08f/c109ae48211709dd500ef97638af308f/c109ae48211709dd500ef97638af308f.db | 17:24 |
| *** joeljwright has left #openstack-swift | 17:24 | |
| zaitcev | /srv/node/b3/containers/197670/08f/c109ae48211709dd500ef97638af308f/c109ae48211709dd500ef97638af308f.db: SQLite 3.x database | 17:24 |
| jaakkos | MooingLemur: was that a raidz system | 17:24 |
| MooingLemur | nope, single disk per pool | 17:25 |
| jaakkos | oh | 17:25 |
| zaitcev | MooingLemur: how do you mange to use ext4? We at RH used to support ext4 as first class citizen for Swift, but gave up when it turned out that its maximum metadata size was too small. | 17:25 |
| MooingLemur | SSDs sliced up to have a log dev and a cache dev per pool | 17:25 |
| MooingLemur | zaitcev: really? What is that maximum size? I stopped using xfs because of the kernel not handling failed disks gracefully and spamming printks until reboot even after umount -fl on the bad mount so that no more filesystem accesses were happening | 17:27 |
| MooingLemur | Granted, this was 4 years ago on early 3.x kernels, but still. | 17:28 |
| zaitcev | MooingLemur: I think it was 64 KB. Note, that was in chunks of something like 1k? IIRC. Swift since removed chunking, so I cannot imagine it working at all. | 17:28 |
| MooingLemur | I mean, I made the inodes larger than default so most xattrs would fit inside the inode | 17:28 |
| *** zul has joined #openstack-swift | 17:29 | |
| MooingLemur | Well, this is news to me | 17:32 |
| MooingLemur | ext4's limit for the entire set of keys and values on a file is apparently the block size of the filesystem (4k in my case) | 17:33 |
| MooingLemur | but I don't think we've run into this | 17:33 |
| *** sgundur_ has quit IRC | 17:34 | |
| *** tqtran has joined #openstack-swift | 17:36 | |
| *** sgundur_ has joined #openstack-swift | 17:38 | |
| MooingLemur | zaitcev: appears the removal of chunking supposedly improved storage efficiency on ext4 (since there were fewer keys, more value can be stored), but otherwise there's no change. I don't think we've ever had the issue of xattrs exceeding 4k here. | 17:42 |
| *** asettle has quit IRC | 17:43 | |
| MooingLemur | but yeah, ext4 wasn't chosen because of performance, it was roughly equivalent to xfs, but xfs was disfavored because of how the kernel handled disappearing disks. | 17:44 |
| MooingLemur | We started our cluster on pretty shoddy hardware, so that kinda colored the design choices :) | 17:44 |
| *** mvk has joined #openstack-swift | 17:45 | |
| openstackgerrit | Charles Hsu proposed openstack/python-swiftclient: Add additional headers for DELETE requests. https://review.openstack.org/372656 | 17:47 |
| *** arch-nemesis has quit IRC | 17:47 | |
| *** acoles is now known as acoles_ | 17:50 | |
| *** oshritf has joined #openstack-swift | 17:54 | |
| *** oshritf_ has joined #openstack-swift | 17:54 | |
| *** oshritf has quit IRC | 17:55 | |
| *** oshritf_ has quit IRC | 17:55 | |
| *** oshritf has joined #openstack-swift | 17:56 | |
| *** oshritf_ has joined #openstack-swift | 17:56 | |
| Renich | MooingLemur: reading... | 17:56 |
| Renich | MooingLemur: I was about to get into the SSD slicing | 17:57 |
| *** manous has quit IRC | 18:14 | |
| *** tsg_ has quit IRC | 18:16 | |
| openstackgerrit | Merged openstack/swift: Bring "egregious range request" comment in line with reality https://review.openstack.org/369060 | 18:20 |
| *** juzuluag has quit IRC | 18:23 | |
| *** manous has joined #openstack-swift | 18:27 | |
| *** sams-gleb has joined #openstack-swift | 18:41 | |
| *** sams-gleb has quit IRC | 18:42 | |
| *** asettle has joined #openstack-swift | 18:43 | |
| *** chsc has joined #openstack-swift | 18:47 | |
| *** asettle has quit IRC | 18:49 | |
| *** oshritf_ has quit IRC | 18:57 | |
| *** oshritf has quit IRC | 18:57 | |
| *** silor has quit IRC | 19:02 | |
| *** manous has quit IRC | 19:04 | |
| *** oshritf has joined #openstack-swift | 19:15 | |
| *** oshritf_ has joined #openstack-swift | 19:15 | |
| *** manous has joined #openstack-swift | 19:16 | |
| *** McMurlock1 has joined #openstack-swift | 19:16 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: add policy info to swift-get-nodes https://review.openstack.org/372716 | 19:19 |
| *** oshritf has quit IRC | 19:22 | |
| *** oshritf_ has quit IRC | 19:22 | |
| *** oshritf has joined #openstack-swift | 19:23 | |
| *** oshritf_ has joined #openstack-swift | 19:23 | |
| *** manous has quit IRC | 19:23 | |
| *** sgundur_ has quit IRC | 19:26 | |
| *** sgundur_ has joined #openstack-swift | 19:30 | |
| *** sgundur_ has quit IRC | 19:34 | |
| *** sgundur_ has joined #openstack-swift | 19:35 | |
| *** manous has joined #openstack-swift | 19:37 | |
| *** McMurlock1 has quit IRC | 19:37 | |
| *** manous has quit IRC | 19:41 | |
| *** asettle has joined #openstack-swift | 19:45 | |
| *** manous has joined #openstack-swift | 19:46 | |
| tqtran | question: if i submitted a patch that tempest is failing on, do i submit an independent patch to tempest with a depends-on tag? or is there a different process? | 19:47 |
| *** asettle has quit IRC | 19:50 | |
| *** asettle has joined #openstack-swift | 19:59 | |
| timburke | tqtran: generally, it depends on whether it's a consistent or intermittent failure | 20:00 |
| *** sgundur_ has quit IRC | 20:03 | |
| *** nikivi has joined #openstack-swift | 20:09 | |
| tqtran | its consistent, im getting it everytime. | 20:10 |
| tqtran | if i patch tempest, it will fail because the swift code isnt in. if i patch swift, tempest fails. i'll dig around the logs a bit more to make sure. but worse case, how do i patch both without one failing since they are co-dependent? | 20:12 |
| timburke | i take it we're talking https://review.openstack.org/#/c/356715/ ? does this mean that the old acl format doesn't work with that patch? | 20:17 |
| patchbot | patch 356715 - swift - Supporting Account ACL in keystoneauth | 20:17 |
| *** asettle has quit IRC | 20:19 | |
| timburke | i don't think this is a problem with tempest. it *is* a sign that swift's func tests need to be more thorough. apparently we're getting a traceback like http://paste.openstack.org/show/581228/ during a /info request | 20:20 |
| openstackgerrit | oshritf proposed openstack/swift: Add process level concurrency to container sync https://review.openstack.org/210099 | 20:21 |
| timburke | posted some comments to help you along | 20:23 |
| tqtran | hm i see, thanks for the comments timburke | 20:24 |
| *** sgundur_ has joined #openstack-swift | 20:24 | |
| timburke | np, happy to help! | 20:24 |
| *** janonymous has quit IRC | 20:28 | |
| *** nikivi has quit IRC | 20:32 | |
| *** oshritf has quit IRC | 20:35 | |
| *** oshritf_ has quit IRC | 20:35 | |
| *** asettle has joined #openstack-swift | 20:38 | |
| *** asettle has quit IRC | 20:43 | |
| *** janonymous has joined #openstack-swift | 20:55 | |
| *** kong has joined #openstack-swift | 21:17 | |
| *** sgundur_ has quit IRC | 21:18 | |
| openstackgerrit | Merged openstack/python-swiftclient: Update help text around container ACL syntax https://review.openstack.org/324739 | 21:24 |
| *** Renich has quit IRC | 21:31 | |
| *** vint_bra has quit IRC | 21:32 | |
| *** Renich has joined #openstack-swift | 21:32 | |
| *** Renich has quit IRC | 21:38 | |
| *** zul has quit IRC | 21:38 | |
| *** zul has joined #openstack-swift | 21:43 | |
| *** asettle has joined #openstack-swift | 21:44 | |
| *** asettle has quit IRC | 21:48 | |
| *** zul has quit IRC | 21:50 | |
| *** janonymous has quit IRC | 21:51 | |
| *** nadeem has quit IRC | 21:51 | |
| *** kencjohnston has quit IRC | 21:51 | |
| *** tdasilva has quit IRC | 21:51 | |
| *** bkeller` has quit IRC | 21:51 | |
| *** blair has quit IRC | 21:51 | |
| *** jaakkos has quit IRC | 21:51 | |
| *** kencjohnston has joined #openstack-swift | 21:51 | |
| *** bkeller` has joined #openstack-swift | 21:51 | |
| *** janonymous has joined #openstack-swift | 21:52 | |
| mattoliverau | Morning | 21:57 |
| openstackgerrit | Tim Burke proposed openstack/swift: Expose encryption status in GET/HEAD/PUT/POST responses https://review.openstack.org/348603 | 22:00 |
| openstackgerrit | Tim Burke proposed openstack/swift: Expose encryption status to client https://review.openstack.org/348604 | 22:00 |
| openstackgerrit | Tim Burke proposed openstack/swift: Include object sysmeta in POST responses https://review.openstack.org/357559 | 22:00 |
| *** tdasilva has joined #openstack-swift | 22:06 | |
| openstackgerrit | Merged openstack/swift: api-ref: clean up account listing examples https://review.openstack.org/366211 | 22:10 |
| *** nadeem has joined #openstack-swift | 22:13 | |
| *** jaakkos has joined #openstack-swift | 22:16 | |
| *** rcernin has quit IRC | 22:27 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Make diskfile verification easier to reason about https://review.openstack.org/372799 | 22:36 |
| *** nadeem has quit IRC | 22:37 | |
| *** blair has joined #openstack-swift | 22:41 | |
| *** asettle has joined #openstack-swift | 22:45 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Clean up api-ref examples https://review.openstack.org/371840 | 22:49 |
| *** asettle has quit IRC | 22:51 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Expose encryption status in GET/HEAD/PUT/POST responses https://review.openstack.org/348603 | 22:56 |
| openstackgerrit | Tim Burke proposed openstack/swift: Expose encryption status to client https://review.openstack.org/348604 | 22:56 |
| *** janonymous has quit IRC | 22:58 | |
| *** fguillot has joined #openstack-swift | 23:12 | |
| *** tgtanya has joined #openstack-swift | 23:14 | |
| *** chsc has quit IRC | 23:36 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift: Add a configurable URL base to staticweb https://review.openstack.org/372809 | 23:37 |
| *** openstackgerrit has quit IRC | 23:42 | |
| *** gopenshaw has quit IRC | 23:43 | |
| *** kei_yama has joined #openstack-swift | 23:44 | |
| *** gopenshaw has joined #openstack-swift | 23:45 | |
| *** Suyi_ has quit IRC | 23:45 | |
| *** asettle has joined #openstack-swift | 23:47 | |
| *** openstackgerrit has joined #openstack-swift | 23:47 | |
| *** asettle has quit IRC | 23:51 | |
| *** openstackgerrit has quit IRC | 23:51 | |
| *** zaitcev has quit IRC | 23:52 | |
| *** torgomatic_ has quit IRC | 23:52 | |
| *** torgomatic has joined #openstack-swift | 23:52 | |
| *** ChanServ sets mode: +v torgomatic | 23:52 | |
| *** zaitcev has joined #openstack-swift | 23:52 | |
| *** ChanServ sets mode: +v zaitcev | 23:52 | |
| *** hogepodge has quit IRC | 23:53 | |
| *** openstackgerrit has joined #openstack-swift | 23:55 | |
| *** hogepodge has joined #openstack-swift | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!