Tuesday, 2016-02-09

*** haomaiwang has quit IRC00:01
*** haomaiwang has joined #openstack-swift00:01
*** dewsday has quit IRC00:02
*** siva_krishnan has joined #openstack-swift00:14
*** daemontool has joined #openstack-swift00:17
*** bill_az has quit IRC00:35
*** arch-nemesis has quit IRC00:39
*** mingyu has quit IRC00:43
*** mingyu has joined #openstack-swift00:44
*** Jeffrey4l has joined #openstack-swift00:44
*** trifon has quit IRC00:45
*** itlinux has joined #openstack-swift00:52
*** Jeffrey4l has quit IRC00:57
*** haomaiwang has quit IRC01:01
*** haomaiwang has joined #openstack-swift01:01
*** garthb has quit IRC01:03
torgomaticanyone know how to run just one functional test with .functests? I tried "./.functests test.functional.test_object:TestObject.test_read_write" like worked with nosetests, but I guess we're using ostestr now (whatever that is) and what i typed doesn't work.01:10
torgomaticalso "./.functests --regex test.functional...." doesn't work either, FWIW01:11
torgomaticokay; got it; it's now "./.functests --no-discover test.functional.test_object.TestObject.test_read_write"; sort of like before, but with extra args and the module:ClassName is now module.ClassName (: to .)01:16
*** dmorita has quit IRC01:22
*** _JZ_ has quit IRC01:23
*** dmorita has joined #openstack-swift01:23
*** pdardeau has joined #openstack-swift01:31
claygtorgomatic: can you not still run them with nosetests?  I was asking timburke how he manages with all this testr crazyness in python-swiftclient - and he was like yeah when I want to run a test I still use nose01:38
*** zhill has quit IRC01:52
*** ChubYann has quit IRC01:56
*** haomaiwang has quit IRC02:01
*** lyrrad has quit IRC02:01
*** haomaiwang has joined #openstack-swift02:01
*** gyee has quit IRC02:06
*** zhill has joined #openstack-swift02:07
torgomaticclayg: that's probably a good idea too; I was just used to using .functests02:08
*** zhill has quit IRC02:11
*** klrmn has quit IRC02:17
*** thurloat is now known as thurloat_isgone02:21
*** asettle is now known as asettle-afk02:23
*** forcebanana has quit IRC02:24
*** mingyu has quit IRC02:35
*** mingyu has joined #openstack-swift02:36
*** dmorita has quit IRC02:42
*** rickyrem1 has joined #openstack-swift02:42
*** dmorita has joined #openstack-swift02:44
*** rickyrem has quit IRC02:44
*** rickyrem1 has quit IRC02:51
*** asettle-afk has quit IRC02:53
*** rickyrem has joined #openstack-swift02:57
*** sanchitmalhotra has joined #openstack-swift03:00
*** haomaiwang has quit IRC03:01
*** haomaiwang has joined #openstack-swift03:01
*** sgundur has left #openstack-swift03:05
*** vinsh has joined #openstack-swift03:11
vinshHey!  When using a dedicated replication network (different port for replicate and non replicate object-servers)  AND making use of "servers_per_port".  Do you need to configure a different port for each disks' replication port the same as the standard port?  http://docs.openstack.org/developer/swift/deployment_guide.html#running-object-servers-per-disk  Is not clear about this combo.03:13
*** klrmn has joined #openstack-swift03:14
vinshSo in that readme example.. "replication_port" might be 6004 6005 6004 6005. compared to "port" being 6000 6001 6000 6001  In that 2 node 2 disk example.  ??03:15
claygvinsh: you do not need to configure a seperate port - you can configure the replication server's port and *not* configure the replication server to use servers per port03:15
claygvinsh: no, i wouldn't configure it it that way - we just use one replication server with multiple workers for all disks (same replication port for all devices on the server)03:16
vinshHey clayg!  I'm finally working on the session feedback you gave me from months ago :)  this is priority number one.  I wonder which has the most benifit for servers_per_port... standard port or replcation_port ?03:16
vinshcool. Understood03:16
claygthere's not a lot of traffic that goes over the object server replication port (the rsync port is unrelated)03:16
vinshI'm trying to get better perf out of my EC cluster... hope this gives me some boost in operations03:17
vinshGood to know, I was thinking this was the case.. but needed to confirm.03:17
claygvinsh: also since it's background work it tpooling of the listdirs in the suffix rehashing causes some stutter in response time's as observed from the remote object-replciator - they're not particuarlly latency sensitive clients03:17
claygwell if you're looking at background reconstruction load in your ec cluster it's going to be quite different - in that it uses ssync as the primary data transport instead of rsync03:19
vinshservers_per_port should be an improvement from just using "workers" as I am now.  I'll roll this out and test03:19
vinshYeah, true.. but it shouldn't have to reconstruct much.. unless data goes missing03:19
claygyes for the standard write path - ec or replicated - servers_per_pork (wokers per disk) is better overall latency and throughput03:19
claygvinsh: correct03:20
vinshI think I abused my cluster with ssbench a bit to hard.  Even after deleting all objects and containers.. the logs on object nodes are a stream of messages similar to: "Feb  9 03:20:09 localhost swift-object-server: Unexpected response: ":ERROR: 500 'ERROR: With :UPDATES: 14 failures to 0 successes'""03:21
vinshIts like its trying to reconstruct or confirm deleted objects still.03:21
vinshBut that's another investigation ;)03:21
vinshI think the only way I solved this in the past was to wipe the account/container drives clean.  stop the ghost churn.03:22
claygweeeeelll, it's maybe some issue with misplaced frags - or some bug related to moving tombstones - there was a few patches for EC reconstructor early on, and few outstanding fixes (all acoles stuff, great stuff, although nothing critical that I'm aware of)03:25
clayg... you might want to do testing with the latest possible release/master you can get ahold of03:26
vinshI'm running 2.5.0-0ubuntu1~cloud0.. so not the most current... but the most current ubuntu has published.03:26
claygyeah 2.5.0 was liberty - there was some good stuff between then and 2.6.0 related to ec i'm *pretty* sure03:27
* vinsh wants 2.6.0 goodness03:27
vinshclayg: Thanks for being such a helpful dude :)03:28
claygvinsh: it's all a front - ask anyone - i'm an asshole03:29
vinsh ! :-D !03:29
*** sanchitmalhotra has quit IRC03:33
*** sanchitmalhotra has joined #openstack-swift03:35
* vinsh gets to building master instead of waiting on cononical's package team.03:36
*** sgundur has joined #openstack-swift03:38
*** sgundur has left #openstack-swift03:38
*** links has joined #openstack-swift03:41
*** forcebanana has joined #openstack-swift03:44
*** asettle-afk has joined #openstack-swift03:46
*** asettle-afk is now known as asettle03:49
*** diazjf has joined #openstack-swift03:54
*** diazjf1 has joined #openstack-swift03:55
*** diazjf has quit IRC03:58
*** diazjf1 has quit IRC03:58
*** haomaiwang has quit IRC04:01
*** haomaiwa_ has joined #openstack-swift04:01
*** rjaiswal has quit IRC04:08
mahaticgood morning04:13
*** asettle has quit IRC04:18
*** asettle has joined #openstack-swift04:19
*** zhill has joined #openstack-swift04:28
openstackgerritMerged openstack/swift: Speed up functional testing  https://review.openstack.org/27746504:31
*** daemontool has quit IRC04:32
*** daemontool has joined #openstack-swift04:32
*** haomaiwa_ has quit IRC05:01
*** haomaiwang has joined #openstack-swift05:01
*** dmorita has quit IRC05:06
mattoliveraumahatic: morning (sorry for late response)05:09
mahaticmattoliverau: :) have a nice evening05:10
*** klrmn has quit IRC05:14
*** SkyRocknRoll has joined #openstack-swift05:17
*** rickyrem has quit IRC05:18
*** daemontool_ has joined #openstack-swift05:21
*** daemontool has quit IRC05:22
*** krishna has joined #openstack-swift05:30
*** dmorita has joined #openstack-swift05:37
*** sanchitmalhotra1 has joined #openstack-swift05:51
*** sanchitmalhotra has quit IRC05:53
*** haomaiwang has quit IRC06:01
*** haomaiwang has joined #openstack-swift06:01
*** trifon has joined #openstack-swift06:02
*** SkyRocknRoll has quit IRC06:07
openstackgerritChristian Berendt proposed openstack/swift: Remove executable flag from sample config file  https://review.openstack.org/27770406:33
*** SkyRocknRoll has joined #openstack-swift06:36
mahaticmattoliverau: oh I got to read the scrollback only now; congrats on the house!06:39
*** haomaiwang has quit IRC07:01
*** haomaiwang has joined #openstack-swift07:01
*** dmorita has quit IRC07:03
*** natarej_ has quit IRC07:21
*** natarej_ has joined #openstack-swift07:21
*** zaitcev has quit IRC07:26
*** mingyu has quit IRC07:32
*** mingyu has joined #openstack-swift07:32
*** dmorita has joined #openstack-swift07:33
*** asettle is now known as asettle-afk07:34
*** cbartz has joined #openstack-swift07:36
*** Zyric has quit IRC07:37
*** dmorita has quit IRC07:37
*** geaaru has joined #openstack-swift07:39
*** SkyRocknRoll has quit IRC07:40
*** Zyric has joined #openstack-swift07:45
*** lyrrad has joined #openstack-swift07:49
*** lyrrad has quit IRC07:54
*** haomaiwang has quit IRC08:01
*** haomaiwa_ has joined #openstack-swift08:01
*** mvk has joined #openstack-swift08:04
openstackgerritMerged openstack/swift: Changed EC backend from jerasure to liberasurecode in examples and docs  https://review.openstack.org/27441608:04
*** Zyric has quit IRC08:04
*** mingyu has quit IRC08:05
*** mingyu has joined #openstack-swift08:05
*** rledisez has joined #openstack-swift08:08
*** Zyric has joined #openstack-swift08:14
*** Jeffrey4l has joined #openstack-swift08:23
*** asettle-afk has quit IRC08:33
*** asettle-afk has joined #openstack-swift08:45
*** mvk has quit IRC08:49
*** asettle-afk has quit IRC08:50
*** acoles_ is now known as acoles08:55
*** bkumar has joined #openstack-swift08:59
*** mingyu has quit IRC09:00
*** haomaiwa_ has quit IRC09:01
*** mingyu has joined #openstack-swift09:01
*** haomaiwang has joined #openstack-swift09:01
acolesmattoliverau: wow! I love the idea of having "Surf Beach" as a line in my address :D09:03
acolesmattoliverau: the balcony looks like it would accommodate a hackathon :P09:04
*** mvk has joined #openstack-swift09:17
acolestorgomatic: clayg fwiw nosetests can still be used to run func tests individually (and collectively) e.g. nosetests ./test/functional/test_object.py:TestObject.test_too_small_x_timestamp09:21
acolesor if you are feeling lucky, nosetests ./test/functional/test_object.py -m too_small09:22
*** joeljwright has joined #openstack-swift09:26
*** ChanServ sets mode: +v joeljwright09:26
*** mvk has quit IRC09:29
*** zhill has quit IRC09:30
*** mvk has joined #openstack-swift09:30
*** SkyRocknRoll has joined #openstack-swift09:33
*** asettle-afk has joined #openstack-swift09:39
*** asettle-afk is now known as asettle09:39
*** daemontool has joined #openstack-swift09:39
*** daemontool_ has quit IRC09:40
*** zhill has joined #openstack-swift09:51
*** zhill has quit IRC09:56
*** takashi has quit IRC10:00
*** haomaiwang has quit IRC10:01
*** haomaiwa_ has joined #openstack-swift10:01
*** Jeffrey4l has quit IRC10:07
*** silor has joined #openstack-swift10:36
*** silor1 has joined #openstack-swift10:41
*** silor has quit IRC10:42
*** silor1 is now known as silor10:42
*** km has quit IRC10:44
*** zhill has joined #openstack-swift10:52
*** zhill has quit IRC10:59
*** haomaiwa_ has quit IRC11:01
*** haomaiwang has joined #openstack-swift11:01
*** mvk has quit IRC11:07
*** mvk has joined #openstack-swift11:08
openstackgerritDonagh McCabe proposed openstack/swift: Added links to API reference and usage guides  https://review.openstack.org/27634211:14
*** jmccarthy has quit IRC11:25
*** jmccarthy has joined #openstack-swift11:27
*** bkumar has quit IRC11:32
*** sanchitmalhotra1 has quit IRC11:34
*** mingyu_ has joined #openstack-swift11:42
*** mingyu has quit IRC11:45
*** zhill has joined #openstack-swift11:55
*** zhill has quit IRC11:59
*** kei_yama has quit IRC11:59
*** haomaiwang has quit IRC12:01
*** haomaiwang has joined #openstack-swift12:01
*** thurloat_isgone is now known as thurloat12:22
*** asettle has quit IRC12:35
*** asettle has joined #openstack-swift12:35
*** asettle has quit IRC12:38
*** openstackgerrit_ has joined #openstack-swift12:39
openstackgerritChristopher Bartz proposed openstack/swift: Adds storage policy option to recon  https://review.openstack.org/27662912:40
*** marcusvrn_ has joined #openstack-swift12:44
*** SkyRocknRoll has quit IRC12:53
*** zhill has joined #openstack-swift12:55
openstackgerritChristian Berendt proposed openstack/swift: Remove executable flag from sample config file  https://review.openstack.org/27770412:57
*** mrmoje has joined #openstack-swift12:59
*** zhill has quit IRC13:00
*** haomaiwang has quit IRC13:01
*** haomaiwang has joined #openstack-swift13:01
*** links has quit IRC13:17
*** bill_az has joined #openstack-swift13:25
*** jordanP has joined #openstack-swift13:41
*** daemontool has quit IRC13:42
*** dmorita has joined #openstack-swift13:46
*** haomaiwang has quit IRC13:46
*** mrmoje has quit IRC13:46
*** haomaiwa_ has joined #openstack-swift13:47
*** mrmoje has joined #openstack-swift13:47
*** openstackgerrit_ has quit IRC13:56
*** zhill has joined #openstack-swift13:56
*** rickyrem has joined #openstack-swift13:57
*** daemontool has joined #openstack-swift13:59
*** zhill has quit IRC14:00
*** haomaiwa_ has quit IRC14:01
*** haomaiwa_ has joined #openstack-swift14:01
*** dmorita has quit IRC14:05
*** petertr7_away is now known as petertr714:05
*** haomaiwa_ has quit IRC14:08
*** andrey-mp has joined #openstack-swift14:11
andrey-mphello. what happens with swift3 project? I'm waiting for review since last August - https://review.openstack.org/#/c/211933/14:16
*** daemontool_ has joined #openstack-swift14:16
*** links has joined #openstack-swift14:16
andrey-mpwho is it possible to review it?14:16
*** daemontool has quit IRC14:17
cbartzandrey-mp: if nobody is answering you here, try to put it on the agenda of the swift meeting: https://wiki.openstack.org/wiki/Meetings/Swift14:18
andrey-mpthank you!14:19
*** andrey-mp has quit IRC14:22
*** rledisez has quit IRC14:28
*** rledisez has joined #openstack-swift14:29
*** rledisez has quit IRC14:30
*** rledisez has joined #openstack-swift14:30
*** portante has quit IRC14:30
*** portante has joined #openstack-swift14:33
*** jordanP has quit IRC14:39
*** tongli has joined #openstack-swift14:40
glangeor try again here later, a lot of people here are on west coast time14:44
*** cbartz has left #openstack-swift14:49
*** amit213 has quit IRC14:51
*** proteusguy_ has quit IRC14:53
*** links has quit IRC14:54
*** zhill has joined #openstack-swift14:57
*** sgundur has joined #openstack-swift15:01
*** zhill has quit IRC15:01
*** _JZ_ has joined #openstack-swift15:03
*** proteusguy_ has joined #openstack-swift15:07
acolesor try #swift3 channel, or look for kota_ in here15:10
*** cdelatte has joined #openstack-swift15:14
*** dmorita has joined #openstack-swift15:21
*** _JZ__ has joined #openstack-swift15:23
*** _JZ_ has quit IRC15:26
*** sgundur has left #openstack-swift15:27
*** petertr7 is now known as petertr7_away15:38
*** mragupat has joined #openstack-swift15:38
*** sgundur has joined #openstack-swift15:45
*** marcusvrn_ has quit IRC15:47
*** dmorita has quit IRC15:56
*** zhill has joined #openstack-swift15:58
*** esker has joined #openstack-swift16:01
*** garthb has joined #openstack-swift16:02
*** arch-nemesis has joined #openstack-swift16:02
*** zhill has quit IRC16:03
*** bjkeller has joined #openstack-swift16:04
*** mvk has quit IRC16:07
*** geaaru has quit IRC16:09
*** daemontool__ has joined #openstack-swift16:12
*** daemontool__ is now known as daemontool16:13
*** daemontool_ has quit IRC16:15
*** petertr7_away is now known as petertr716:18
openstackgerritChaozhe Chen(ccz) proposed openstack/python-swiftclient: Use "# noqa" instead of "#flake8: noqa"  https://review.openstack.org/27792716:19
*** diazjf has joined #openstack-swift16:27
*** dmorita has joined #openstack-swift16:28
*** petertr7 is now known as petertr7_away16:35
*** petertr7_away is now known as petertr716:38
*** nadeem has joined #openstack-swift16:40
openstackgerritAlistair Coles proposed openstack/swift: Merge branch 'master' into feature/crypto  https://review.openstack.org/27795016:42
acolesjrichli: mahatic ^^ requirements.txt changed so we need this to pass gate checks16:43
jrichliacoles: ok, thx16:44
mahaticacoles: thanks16:44
notmynamegood morning16:44
acolesgood morning16:45
*** zaitcev has joined #openstack-swift16:45
*** ChanServ sets mode: +v zaitcev16:45
jrichligood morning16:45
openstackgerritMingyu Li proposed openstack/swift: Order the devices by (region, zone, ip, device) in the output of swift-ring-builder  https://review.openstack.org/27795616:49
*** esker has quit IRC16:49
*** zhill has joined #openstack-swift16:58
*** mragupat has quit IRC17:00
*** mragupat has joined #openstack-swift17:01
*** d0ugal has quit IRC17:02
*** zhill has quit IRC17:03
*** daemontool has quit IRC17:06
*** gyee has joined #openstack-swift17:12
*** lyrrad has joined #openstack-swift17:17
*** bjkeller has left #openstack-swift17:21
*** bjkeller1 has joined #openstack-swift17:21
*** asettle has joined #openstack-swift17:21
*** venkat has joined #openstack-swift17:24
venkatHi all17:25
-openstackstatus- NOTICE: Gerrit is restarting now, to alleviate current performance impact and WebUI errors.17:25
venkatsometimes I am getting the following error for memcache17:25
venkatSTDERR: ERROR:root:Timeout getting a connection to memcached:
venkatwhat is the reason17:25
venkatmax_connections = 217:25
*** asettle has quit IRC17:26
*** rickyrem has quit IRC17:27
*** mragupat has quit IRC17:32
*** dmorita has quit IRC17:32
*** mragupat has joined #openstack-swift17:33
venkatHi someone can answer17:34
*** venkat has quit IRC17:35
*** mragupat has quit IRC17:36
*** mragupat has joined #openstack-swift17:37
jrichlivenkat: have you looked at why your connection might be timing out?  maybe some details on that investigation would help.17:40
*** rledisez has quit IRC17:41
*** klrmn has joined #openstack-swift17:47
*** ntt has quit IRC17:50
*** silor has quit IRC17:50
*** jordanP has joined #openstack-swift17:55
*** joeljwright has quit IRC18:00
*** dmorita has joined #openstack-swift18:01
*** jordanP has quit IRC18:03
*** bjkeller1 has quit IRC18:04
*** bjkeller has joined #openstack-swift18:05
*** mrmoje has quit IRC18:08
*** zhill has joined #openstack-swift18:08
*** zhill has quit IRC18:08
*** manous_ has joined #openstack-swift18:09
manous_how can i get API KEY for use netdrive  ?18:09
openstackgerritAlistair Coles proposed openstack/swift: Decrypt multipart mime doc  https://review.openstack.org/24837718:11
*** zhill_desktop has joined #openstack-swift18:13
*** asettle has joined #openstack-swift18:17
*** petertr7 is now known as petertr7_away18:18
*** _JZ__ has quit IRC18:20
*** asettle has quit IRC18:21
*** delattec has joined #openstack-swift18:23
*** daemontool has joined #openstack-swift18:25
*** cdelatte has quit IRC18:26
*** acoles is now known as acoles_18:29
*** _JZ_ has joined #openstack-swift18:30
*** esker has joined #openstack-swift18:30
*** zhill has joined #openstack-swift18:32
*** mragupat_ has joined #openstack-swift18:32
*** bjkeller has quit IRC18:32
*** mragupat has quit IRC18:36
*** zhill_desktop has quit IRC18:39
*** ChubYann has joined #openstack-swift18:44
*** delatte has joined #openstack-swift18:48
*** delattec has quit IRC18:51
*** ober37 has joined #openstack-swift18:57
siva_krishnanHi all! just  had one quick question on ACL's. Can admin user grant  read-write or full access to the non-existent user  ?19:06
jrichlisummit voting is open19:07
claygjrichli: oh no!19:08
claygsiva_krishnan: sure!19:08
*** daemontool_ has joined #openstack-swift19:10
*** daemontool__ has joined #openstack-swift19:11
pdardeaucuriosity is killing me. what is the non-existent user? a placeholder of some kind?19:11
*** daemontool has quit IRC19:12
klrmnpdardeau: i would guess a user that doesn't exist yet…or at least that swift hasn't created an account for yet.19:12
siva_krishnan+1 klrmn19:14
*** daemontool_ has quit IRC19:14
*** esker has quit IRC19:16
*** esker has joined #openstack-swift19:17
*** esker has quit IRC19:19
*** forcebanana has quit IRC19:23
siva_krishnanclayg: can you please let me know if it would be possible for an admin user to grant read-write  access to user who is not existing in the system19:33
notmynamedidn't he already answer "yes"?19:33
*** ChubYann has quit IRC19:33
*** esker has joined #openstack-swift19:34
*** mingyu_ has quit IRC19:34
*** esker has joined #openstack-swift19:34
notmynamesiva_krishnan: there's no validation against an identity system when you set ACLs. so you can set it to any arbitrary string. of course, if it does end up matching something in the identity system, then you get what you expect (ie ACL functionality)19:35
notmynameso you can set the ACL values to anything you want. doesn't have to be a real user that exists right now19:35
*** mingyu has joined #openstack-swift19:35
pdardeaunotmyname: thanks for elaborating19:35
siva_krishnannotmyname: thanks for the info.19:37
*** can8dnSix has joined #openstack-swift19:41
*** ChubYann has joined #openstack-swift19:44
*** forcebanana has joined #openstack-swift19:45
*** bjkeller has joined #openstack-swift19:50
*** esker has quit IRC19:56
*** can8dnSix has quit IRC19:59
*** ober37 has left #openstack-swift20:06
*** ober37 has joined #openstack-swift20:06
*** ober37 has left #openstack-swift20:06
*** petertr7_away is now known as petertr720:07
*** rcernin has joined #openstack-swift20:08
*** manous_ has quit IRC20:14
jrichlinotmyname: should I indicate "docimpact" on the commit message for patch 252791?20:33
patchbotjrichli: https://review.openstack.org/#/c/252791/ - swift - Change POST-as-COPY to preserve the DLO manifest c...20:33
notmynamejrichli: that is a good question20:40
notmynamejrichli: the definition and what happens with docimpact has been changing lately20:40
notmynamejrichli: why do you think it should or should not have docimpact?20:40
jrichlinotmyname: it includes doc changes.  in particular, it is clarifying that a client must provide the x-object-manifest when POSTing to a dlo manifest.  This behavior was something that had to be discussed on the related bug report.20:43
jrichlinotmyname: but it doesn't seem like all doc changes require the docimpact, so i don't know where the line is drawn.20:43
notmynameok, thanks. I asked because it is and has been so confusing as to when to use it! it's good to hear what other people think it's supposed to mean20:45
notmynamein general, docimpact is supposed to be for when a patch changes something that is documented in the openstack-manuals repo20:47
notmynamewhich raises the question, what's in the openstack-manuals repo? and I don't know ;-)20:47
notmynamehowever, docimpact is likely changing so that it will open a bug on swift itself saying something like "this patch had a docimpact tag"20:48
notmynameand much of the openstack-manuals content is moving into our own in-tree repo anyway20:49
notmyname(like at least some of the API docs)20:49
notmynameso all that being said, basically adding docimapct has one real result: asettle will come ask what to do with it20:49
jrichlilol.  Well, asettle could point me to what is currently in the openstack-manuals, right?20:50
notmynamein theory. she and I talked a bunch about this last week in australia20:51
jrichlii would think that is straightforward : look at the repo in github?  not so much?20:51
notmynamejrichli: it's the stuff hosted at http://docs.openstack.org20:52
jrichliok, so thats what i frequently look at :-)20:52
*** esker has joined #openstack-swift20:53
jrichlinotmyname: so I would just make a proposal for how to add this specification into that format? and asettle can decide if it doesn't belong at that level I suppose20:55
notmynamejrichli: IMO we'll probably stop using docimpact. but for now there's not really any harm in adding it.20:55
notmynameand asettle hangs out in here. it's just before 8am for her now, so she'll probably be around in 90minutes or so20:55
jrichliok.  thanks!20:56
*** mrmoje has joined #openstack-swift20:59
*** esker has quit IRC20:59
notmynameredbo_: do you remember why swift's timestamp resolution is decimillisecond?21:00
openstackgerritMichael Barton proposed openstack/swift: go: GREP middleware  https://review.openstack.org/21738121:00
redbo_notmyname: nope21:01
*** redbo_ is now known as redbo21:01
*** ChanServ sets mode: +v redbo21:01
*** daemontool has joined #openstack-swift21:02
*** daemontool__ has quit IRC21:02
openstackgerritMichael Barton proposed openstack/swift: go: GREP middleware  https://review.openstack.org/21738121:03
*** _JZ_ has quit IRC21:04
openstackgerritClay Gerrard proposed openstack/python-swiftclient: more tests for pseudo/dir  https://review.openstack.org/27809021:15
mrmojeclayg, timburke: muchos gracias!21:17
glangedecimilliseconds == microseconds, I think21:19
glangeno, that's not right21:21
*** mrmoje has quit IRC21:22
glangeit's also decideciseconds21:22
glangeer, centicentiseconds21:23
glangelike a second centipede21:23
*** mrmoje has joined #openstack-swift21:24
*** daemontool_ has joined #openstack-swift21:24
*** daemontool has quit IRC21:27
*** mrmoje has quit IRC21:32
*** mrmoje has joined #openstack-swift21:34
*** mrmoje has quit IRC21:40
pdardeaugood morning mattoliverau21:43
vinshHey all.. when using "servers_per_port" does the "bind_port" setting matter?21:50
vinshdo you just set bind_port to every port you are using for each disk?21:50
notmynamevinsh: bind_port matters, but worker doesn't21:51
vinshSuppose I have 30 disks and thus 30 different ports for "servers_per_port".  what do I choose for "bind_port" ?21:51
notmynamevinsh: and yes, you'd end up with one object server config for each drive (each with a different bind_port setting). and then each drive in the ring would have a different port21:51
vinshbind_port appears only under the DEFAULT section from what I can tell21:53
openstackgerritShashirekha Gundur proposed openstack/swift: change default ports for servers  https://review.openstack.org/27484021:53
*** petertr7 is now known as petertr7_away21:55
*** peyton has quit IRC21:58
claygnotmyname: I always thought it had to do with float precision of six-digit literal flots!21:59
vinshnotmyname: Can you describe the config section a bit more?21:59
openstackgerritMerged openstack/swift: Insert versioned_writes in correct pipeline position  https://review.openstack.org/27307321:59
vinshIf I start an object server with servers_per_port set.  it starts x number of servers per port defined for that ip in the ring.21:59
notmynameclayg: yeah. or maybe some platforms don't give microseconds?21:59
vinshI can't then start another object server because that port is now in use.22:00
claygnotmyname: some python floats can be "equal" even tho they have different first six digits (1455054922.062829 == 1455054922.0628289)22:00
claygnotmyname: ^ i don't think that's true for the first five digits?22:00
*** esker has joined #openstack-swift22:00
notmynameah, interesting22:00
notmynameby which I mean "that's terrible"22:01
notmynamevinsh: suppose you have 3 drives A, B, C22:01
notmynamevinsh: and you have servers_per_port set to 222:01
claygnotmyname: i mean if you're giving 10 digits to the before the decimal you ony have so much precision afterwards - and the answer of how "how much" was 5 digits22:01
claygat least that's always what I told myself22:02
notmynamevinsh: so you'd end up with 3 config files. each with a different bind_port. eg 10001, 10002, 1000322:02
vinshso object-server/1.conf 2.conf 3.conf maybe.22:02
notmynamevinsh: so you start each of these object servers, and they will each make 2 workers22:02
vinshah. but they create 622:02
vinshnot just 222:02
clayg6 object-server conf?  that's baller22:03
notmynameand in your ring, you'll add the 3 drives and importantly add them with their specific ports22:03
vinshIn notmyname's example. when I start the first server. it starts 6. not just 2.22:03
notmynamevinsh: create 6? ie 3 servers each with 2 child?22:03
notmynamevinsh: ah, that's not right. you need to make sure the ring is updated22:03
notmynamevinsh: with different ports22:04
notmynameso that each will only start workers for it's "local" drives. where local == drives with the same ip and port as the loaded config22:04
*** peyton has joined #openstack-swift22:05
*** esker has quit IRC22:05
notmynameyou repeat ports in there22:05
notmynameoh wait22:05
notmynamesorry. you're right22:05
*** ober37 has joined #openstack-swift22:05
vinshI PRONMISE to put up a doc review/update once I can make more sense of how this is done :)22:06
*** ober37 has quit IRC22:06
notmynameI've set it up myself. it's just been a while :-)22:06
notmynamewhat's your workers_per_port setting?22:07
vinshthats object-server/1.conf for example.22:07
timburkenotmyname: clayg: oh, of course! we've only got a 52-bit significand in a 64-bit float and 1e5 < (1<<53) / 1e10 < 1e622:08
vinshsince that node has 4 drives.  starting this file will start 8 workers total.22:08
vinshbut that file only has a bind_port for 1/4 ports used in the ring.22:08
*** trifon has quit IRC22:09
* notmyname wonders if swifterdarrell is busy22:09
openstackgerritBen Keller proposed openstack/swift: Headers now use a case incensitive dictionary  https://review.openstack.org/27490622:09
vinshOh. I can omit bing_port from 1.conf and servers start no problem22:11
vinshWhich makes sense I guess.. since thats what servers_per_port is all about.. using the ring to spawn wokers per port per local ip22:11
notmynameyou shouldn't be able to omit bind_port. it's been required for a while22:14
notmynamevinsh: http://docs.openstack.org/developer/swift/deployment_guide.html#running-object-servers-per-disk22:15
vinshwell. I can omit it and it works22:16
vinshthat deployment guide makes no mention of different config sections per server port22:16
vinshI have read it 10 times over :)22:16
*** diazjf has quit IRC22:20
*** asettle has joined #openstack-swift22:24
notmynameasettle: good morning. jrichli had a question for you earlier22:24
notmynameabout docimpact22:24
asettlenotmyname: what can I help with?22:24
jrichliasettle: I was thinking we might want to make doc updates for patch 25279122:25
patchbotjrichli: https://review.openstack.org/#/c/252791/ - swift - Change POST-as-COPY to preserve the DLO manifest c...22:25
asettleYep, let me look22:26
asettlejrichli: so you're looking at getting this content moved into the openstack-manuals side of things?22:27
jrichliasettle: once this patch would be merged, I would think we could add a statement about having to specify the "x-object-manifest" for a DLO manifest in the Object POST area.22:28
jrichliyes, into openstack-manuals22:29
asettleGive meeee a minuteeee :) if you want, I am happy to be assigned to this patch so I can monitor when it's merged. Let me just double check we actually have DLO info in the manuals.22:30
asettleIf not, would you like that to be added into the os-manuals?22:30
asettleThe whole thing, that is.22:30
jrichliasettle: I dont feel that is my call ... I know that DLO is at least mentioned in some spots now, such as "the manifest is returned in the X-Object-Manifest response header for dynamic large objects"22:32
asettleYep, okay. Fair :) sorry, throwing out ideas... I literally just woke up. Haha.22:33
jrichliasettle: no worries.  good morning!22:33
asettleHaha thank you! :)22:34
*** esker has joined #openstack-swift22:34
asettleOkay, I'll look into this for you. If possible jrichli could you please file a bug in OS manuals, tag swift and assign me? Explain what it is you're looking to add and I'll ensure that information gets added in :)22:35
jrichliok, will do22:35
notmynamevinsh: ok, after some code diving...22:39
asettlejrichli: thank you kindly :)22:40
notmynamevinsh: you want one config file. set servers_per_port to whatever. make sure you have a different port for each drive in the ring. it would seem that bind_port should be removed (instead of it being ignored)22:40
notmynamevinsh: so yeah. what you got to, I think :-)22:41
jrichliasettle notmyname: thank you22:41
vinshPerfect, I concur with your read notmyname.  Thank's for looking at this with me!22:41
notmynamevinsh: good luck! (I've seen it improve performance in clusters)22:41
vinshcranking up our EC cluster by using server_per_port.22:41
vinshvideo team wants more I/O. I'll give them more I/O ;)22:42
vinshI wonder servers_per_port makes sense for account/container rings. probably not as much.22:42
*** esker has quit IRC22:46
torgomaticvinsh: might still be worth doing. the point of it is to distribute the blocking file IO among multiple processes so there can be lots of concurrent IOs going. sqlite locking adds some overhead, but there's still disk IO going on22:49
vinshSo account/container servers will use this feature?22:49
vinshI think I'll give this a try!22:49
claygvinsh: not until someone writes that code - I don't think it'll work as-is - object-server only22:50
vinshd'oh :D22:50
* vinsh wants turbo everything.22:50
vinshOne last question for the day... when using EC.. has the ccount/container ring been setup using EC or is it strictly the object ring?22:52
*** esker has joined #openstack-swift22:52
vinshoh nevermind.22:53
vinshthere is no "swift-account|container-reconstructor"22:53
notmynameright. accounts and containers are only replicated. no separate storage policies there22:53
torgomaticoh. heh. forgot about that little detail.22:56
*** tongli has quit IRC22:56
*** rcernin has quit IRC22:58
*** daemontool__ has joined #openstack-swift23:02
*** daemontool_ has quit IRC23:04
jrichliasettle: I was only allowed to add you as a subscriber to the bug, but not allowed to assign to anyone outside of my IBM Stackers team.  unless i am doing it wrong ;-)23:05
asettlejrichli: haha oh right! OKay. let me go take a look :)23:06
asettleExcellente, fixed that :)23:07
*** bjkeller has left #openstack-swift23:07
jrichliasettle: thx!23:08
asettleNo problem :) if i'm not online, feel free to just subscribe me to bug reports - i'll keep an eye out for them.23:09
asettleIf not, just email me23:09
asettleI'm Australia BNE... which is like... something smoething +10 (it's early, don't judge me)23:09
asettleNext week I'll be UK though, so I'll be online a little more often for the US23:09
jrichliasettle: great!  I already had to look you up on linkedin to find out whether to use the id/email you had for rackspace or redhat :-)23:10
asettleHAHA OHHHH yes apologies. Racker ;)23:11
asettleI really should make that clearer somewhere.23:11
jrichliasettle: np.  Have fun in the UK!23:11
asettleWill do! Thank you :)23:11
*** Jeffrey4l has joined #openstack-swift23:15
notmynamejrichli: also, asettle will join us for the hackathon23:16
asettleOh yes, I'll be there.23:16
*** daemontool_ has joined #openstack-swift23:20
*** daemontool__ has quit IRC23:22
*** daemontool_ has quit IRC23:28
*** kei_yama has joined #openstack-swift23:28
*** daemontool has joined #openstack-swift23:29
torgomaticthe best part about centimillisecond precision is that it'll be too much someday23:31
*** ho_ has quit IRC23:44
*** ho_ has joined #openstack-swift23:49
ho_good morning!23:50
*** nadeem has quit IRC23:51
mattoliverauho_: morning23:52
ho_mattoliverau: morning!23:54
*** mragupat_ has quit IRC23:57
*** peyton has quit IRC23:58
*** peyton has joined #openstack-swift23:59

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!