Friday, 2015-12-18

*** chlong has joined #openstack-swift00:03
*** nadeem has joined #openstack-swift00:04
*** lyrrad has joined #openstack-swift00:04
*** alejandrito has quit IRC00:08
hopdardeau: morning!00:08
then3rdWoot, evyrything is running Liberty.00:10
*** zaitcev has quit IRC00:11
notmynamethen3rd: nice!00:11
notmynamethen3rd: no client impact?00:11
then3rdI had one blip where both backends were down, but haproxy let me know right away. It showed (if i recall) a layer7 timeout, even though the swift-proxy was running.00:13
then3rdI'm sure it would've been fine if there were 3+00:14
*** lpabon has quit IRC00:14
then3rdHmm. It seems there may be a problem. Something is running slow.. the healthcheck takes a while to load, so haproxy is dropping it.00:33
*** rohit_ has quit IRC00:33
notmynamewhat if you curl directly to a proxy ip/port and ask for the heathcheck?00:34
*** rohit_ has joined #openstack-swift00:34
notmynameeg `time curl -i http[s]://proxyIP:port/healthcheck`00:34
notmynameerr...almost definitely would be http there00:34
*** daddyjoseph97 has quit IRC00:35
*** SkyRocknRoll has joined #openstack-swift00:35
then3rdWell, I didn't try curl while it was misbehaving, but I hit one of the proxies directly from my browser.00:36
then3rdLets see if it happens again.00:36
kota_good morning00:43
kota_gmmaha: i thing we don't have to remove that. just "sudo pip install -U pip" via OS maintained python-pip. That way worked well in my clean Ubuntu box in AWS.00:45
then3rdOkay, it was dropped again. Curl just stalled the first few times, eventually it returned a 200 after 2+ seconds00:48
*** blmartin has quit IRC00:58
*** nakagawamsa has joined #openstack-swift01:01
*** lyrrad has quit IRC01:01
*** daemontool has quit IRC01:08
*** darrenc_afk is now known as darrenc01:13
then3rdThat's it. I should've checked my logs a long time ago.01:14
then3rdrsyslog is messed up and not ending up in logstash for one of the proxies, but the one that is reporting...01:15
*** dmorita has quit IRC01:18
*** dmorita has joined #openstack-swift01:20
then3rdSTDERR: ERROR:oslo.messaging._drivers.impl_rabbit:AMQP server on is unreachable: [Errno 111] ECONNREFUSED.... over and over and over. It's because I kept celiometer enabled, but didn't setup the rabbitmq options yet. Just dropped it from the pipeline.01:29
*** dmorita has quit IRC01:32
*** dmorita has joined #openstack-swift01:38
homattoliverau: thanks for review. as for the monkey patch, base instance is created in every method so i think we don't need to clean up. is it better to use mock in this case?01:39
mattoliverauho: that's true (re: cleaning up). So I guess we don't have to worry about it getting dirty. I figure we are importing patch, so we should be consistant and use it.. but maybe it's more of a NIT.01:48
homattoliverau: thanks. I will follow our history :-)01:51
*** nadeem has quit IRC01:55
*** haomaiwa_ has joined #openstack-swift01:56
openstackgerritCharles Hsu proposed openstack/python-swiftclient: Add another content-type for directory.
*** haomaiwa_ has quit IRC02:01
*** janonymous_ has quit IRC02:01
*** haomaiwang has joined #openstack-swift02:03
openstackgerritHisashi Osanai proposed openstack/swift: Fix duplication for headers in Access-Control-Expose-Headers
*** Jeffrey4l has joined #openstack-swift02:13
kota_recent gerrit UI change still makes me to confuse the patch chain dependencies (or conflict with) :/02:17
*** badari_ has quit IRC02:21
mattoliveraukota_: yeah I find that too02:22
kota_mattoliverau: i commented 2 duplicated patches (just i was thinking they are depending) :/02:23
kota_and then, i noticed the duplication by notmyname comment02:24
kota_my bad02:24
openstackgerritMerged openstack/swift: Fix func test --until-failure and --no-discover options
*** kragniz has quit IRC02:41
*** kragniz has joined #openstack-swift02:46
*** haomaiwang has quit IRC03:01
*** haomaiwang has joined #openstack-swift03:01
*** lcurtis has quit IRC03:01
*** JohnnyCI has joined #openstack-swift03:02
*** JohnnyCI has left #openstack-swift03:02
*** sanchitmalhotra has joined #openstack-swift03:02
openstackgerritMerged openstack/swift: Pass HTTP_REFERER down to subrequests
*** rminmin has joined #openstack-swift03:10
*** Jeffrey4l has quit IRC03:14
*** links has joined #openstack-swift03:14
*** nadeem has joined #openstack-swift03:16
*** venkat has joined #openstack-swift03:19
*** garthb has joined #openstack-swift03:23
openstackgerritHisashi Osanai proposed openstack/swift: Add X-Delete-After and X-Delete-At functional test
*** Jeffrey4l has joined #openstack-swift03:27
*** Jeffrey4l has quit IRC03:33
*** gyee has quit IRC03:40
*** Jeffrey4l has joined #openstack-swift03:40
*** janonymous has quit IRC03:43
*** SkyRocknRoll has quit IRC03:46
*** yatin has joined #openstack-swift03:49
*** Jeffrey4l__ has joined #openstack-swift03:57
*** Jeffrey4l has quit IRC03:58
*** Jeffrey4l__ is now known as Jeffrey4l03:58
*** SkyRocknRoll has joined #openstack-swift04:00
*** haomaiwang has quit IRC04:01
*** haomaiwang has joined #openstack-swift04:01
*** suresh has joined #openstack-swift04:03
sureshIs there any way to add meta quota bytes to an swift account with out adding reselleradmin role. Please someone help.04:06
*** haomaiwang has quit IRC04:15
*** haomaiwang has joined #openstack-swift04:16
openstackgerritBing Hu proposed openstack/python-swiftclient: ecnode auth token in swiftclient log message Closes-bug:#1516692
*** lpabon has joined #openstack-swift04:21
*** proteusguy_ has quit IRC04:22
*** cebruns has quit IRC04:27
*** albertom has quit IRC04:28
*** cebruns has joined #openstack-swift04:28
*** albertom has joined #openstack-swift04:29
wbhubernotmyname: kota_: left a comment for u 2 for patch 24821004:32
patchbotwbhuber: - Update Erasure Coding Overview doc to remove Beta ...04:32
*** proteusguy_ has joined #openstack-swift04:34
*** janonymous has joined #openstack-swift04:36
*** takashi has joined #openstack-swift04:46
portantethen3rd: are you still have problems with rsyslog?04:47
sureshIs there any way to add meta quota bytes to an swift account with out adding reselleradmin role. Please someone help.04:47
notmynamesuresh: no04:48
*** badari has joined #openstack-swift04:49
*** david-lyle has quit IRC04:50
sureshnotmyname: tnqq04:52
*** esker has quit IRC04:52
sureshnotmyname: if i added reseller admin role to an account it is accepting to upload more than what i set the size in quota-bytes04:54
notmynamesuresh: the quotas are soft quotas, not hard limits.
openstackgerritBing Hu proposed openstack/swift: catch lock fail exception in container updater Closes-bug: #1400939
openstackbug 1400939 in OpenStack Object Storage (swift) "LockTimeout exception not caught in container updater process_container" [Undecided,New]
notmynamewow. finding different patch sets (and which one you are currently looking at) is a *lot* harder in the new gerrit UI04:57
notmyname(there's a tiny "pull down" menu in the top right)04:58
*** klrmn has quit IRC04:59
*** haomaiwang has quit IRC05:01
*** haomaiwang has joined #openstack-swift05:01
sureshnotmyname: I understanded this but what i am asking is  if  i assigned reselleradmin role to user it is accepting to upload more. But when i am removing the role it is not allowing to upload more than the quota-bytes05:03
notmynameZyric: still here?05:09
Zyricnotmyname: Hello, yes I'm here05:10
notmynameZyric: I'm (finally) looking at patch 24197805:10
patchbotnotmyname: - Add support for storage policies to have more than...05:10
notmynameZyric: the logic in change_primary_name() worries. me. can you walk me through it?05:11
notmynameZyric: what happens if the policy only has one name05:11
Zyricnotmyname: If a policy only has one name, and you enter a name that doesn't exist, it adds it as the new primary name.05:13
ZyricIf it's a valid name05:13
sureshIf  i assigned reselleradmin role to user it is accepting to upload more. But when i am removing the role it is not allowing to upload more than the quota-bytes05:14
sureshplease anyone hlep05:14
notmynameZyric: but it calls remove_name() on the old one first. doesn't that error if it's the only name? you haven't added the new name to the list yet05:14
Zyricnotmyname: Ah, you're right, it would prompt an error in that case.05:16
mattoliverausuresh: assigning a reseller role to a user gives them permissions to set ACLs on the account etc. If they have the role or not it doesn't change how they are affected by any account quotas. If there is an account quota then there is a soft limit applied. (thats my understanding anyway)05:18
mattoliveraunotmyname: you can use 'n' and 'p' to move through the patchsets as well, not that it really helps cause you need to look at the top to see what patchset your up to.05:19
mattoliverausuresh: tho I could be wrong, but would need to go look at the code05:20
notmynameI'm looking at the code now05:20
Zyricnotmyname: Is there anything else you can see that needs fixing? I'm working on an account version of patch 212824 at the moment and not sure how to go about switching back to work on the multiple names patch without breaking things.05:22
patchbotZyric: - Let operators add watchers to object audit05:22
sureshmattoliverau: thanq05:23
*** lpabon has quit IRC05:24
notmynameZyric: that's all I saw. I was looking at a diff from patch set 7 (instead of the whole thing). I liked patch set 7, so I was taking the shortcut to first look at just those changes05:25
notmynameZyric: so switching between concurrent things you're working on...05:25
notmynameZyric: judicious use of git ;-)05:25
notmynameZyric: make sure you're using different branches (ie not just patching master) for each thing you're doing. and for little things, `git stash` is your friend05:26
notmynameothers may have different suggestions for managing workflow05:26
mahaticnope, git stash and git apply are my friends too05:26
mattoliveraugit branch, git stash, and sometimes pushing up to github clone so I can access what I'm working on on swift clusters in the cloud05:28
ZyricThanks everyone, I'll try some git tricks then. Will update the multiple names patch change_primary_name() stuff after I come back from dinner :)05:29
notmynamemahatic: `git stash pop` is a good alternative to `git stash apply` to keep the stash cleaner05:30
mattoliverauYou can always git commit and then flatten with git rebase -i, or continually commit --amend. But that is only required if working on something more complex then a smaller change.05:31
mahaticnotmyname: ah I see. True! thanks05:31
notmynamealso, I think "alias.amend=commit --amend" is a really really useful git alias to have05:32
notmynameso I can `git amend -a -m foobar` and it works. with `git commit --amend` it's too easy to forget one "-" and to `git commit -amend` and have everything added to the current commit with the commit message of "end"05:33
*** ppai has joined #openstack-swift05:37
notmynamesuresh: I tried to duplicate what you are describing. I suspect you're getting the larger amount allowed because of the account info that is cached05:37
notmynamesuresh: if there is account info recently cached in the system, swift won't go all the way to disk to refresh it every time it's requested05:38
notmynameand that's where quota info is kept05:38
notmynamesuresh: hmm...maybe not05:39
notmynamesuresh: oh, yeah. now i see it in the code. the reseller admin bypasses quotas05:40
notmynamesuresh: which makes sense, since the reseller can set the quota to anything, there's no point in blocking requests based on a quota05:41
notmynameZyric: cool that you're looking at the account auditor hooks. torgomatic has been out sick this week, but definitely find him online. he'll offer good input on it05:43
*** badari has quit IRC05:46
mattoliveraucool, well there you go then, resellers are magical :)05:51
*** SkyRocknRoll has quit IRC05:53
*** haomaiwang has quit IRC06:01
*** haomaiwang has joined #openstack-swift06:01
*** takashi has quit IRC06:04
*** serverascode has quit IRC06:05
*** nottrobin has quit IRC06:06
*** nottrobin has joined #openstack-swift06:11
*** serverascode has joined #openstack-swift06:14
*** david-lyle has joined #openstack-swift06:22
*** trifon has joined #openstack-swift06:22
*** trifon has quit IRC06:32
*** haomaiw__ has joined #openstack-swift06:32
*** haomaiwang has quit IRC06:34
*** trifon has joined #openstack-swift06:36
*** takashi has joined #openstack-swift06:38
*** dmorita has quit IRC06:45
*** dmorita has joined #openstack-swift06:45
*** SkyRocknRoll has joined #openstack-swift06:46
*** nadeem has quit IRC06:48
*** dmorita has quit IRC06:50
*** zhill has quit IRC07:00
*** haomaiw__ has quit IRC07:01
*** haomaiwang has joined #openstack-swift07:01
*** SkyRocknRoll has quit IRC07:02
venkatHi all,07:03
venkathow to run a single testcase-file in swift?07:03
*** dmorita has joined #openstack-swift07:05
*** esker has joined #openstack-swift07:07
*** zhill has joined #openstack-swift07:09
*** dmorita has quit IRC07:10
*** esker has quit IRC07:12
*** mac_ified has quit IRC07:14
*** SkyRocknRoll has joined #openstack-swift07:14
*** suresh has quit IRC07:15
mattoliverauvenkat: nosetests <path to file>07:22
*** ppai has quit IRC07:23
*** nfdeswqa has joined #openstack-swift07:28
venkatok, thanks07:29
venkatis it possible to run these tests from eclipse directly?07:29
nttHi, someone can give me some advises for buying storage servers for swift? (vendor, model, ecc....)07:35
*** chlong has quit IRC07:37
*** ppai has joined #openstack-swift07:37
openstackgerritKota Tsuyuzaki proposed openstack/swift: Sleep enough for trampoling
*** rcernin has joined #openstack-swift07:41
*** garthb has quit IRC07:46
*** ppai has quit IRC07:52
*** rminmin has quit IRC07:55
*** arnox has joined #openstack-swift07:58
*** haomaiwang has quit IRC08:01
*** haomaiwang has joined #openstack-swift08:01
*** ppai has joined #openstack-swift08:05
*** rledisez has joined #openstack-swift08:07
*** nfdeswqa has quit IRC08:11
yatinI'm getting Insufficient storage error when creating object08:11
yatinI checked all configuration params for object-server conf08:12
yatinis there minimum size of disk, as i'm using saio with 4 disks of 2GB each08:13
yatinable to create container under test account but not able to put object08:14
yatincurl -i -XPUT -H "X-Auth-Token: AUTH_tk62a1cc8c872243e2b568cab13bd8a75c" http://swift:8080/v1/AUTH_test/c/foo --data-binary 123408:14
*** d0ugal has joined #openstack-swift08:14
yatini'm using tempAuth08:15
kota_yatin: Are all disks mounted?08:15
yatinkota_: yes,,, df -h shows mounted08:15
kota_oh, well. if all disks are used for account/container/object, should be same behavior.08:16
yatinkota_: account-server/1.conf , devices = /srv/node/sdb08:16
yatinkota_: the same is path in fstab, so disk is mounted08:17
kota_how many replicas do you set?08:17
yatinkota_: ring i'm building as swift-ring-builder object.builder add r1z1- 108:17
yatinkota_: 308:17
takashiyatin_: Did you check the permission of /srv/node/sdb?08:17
yatinkota_: swift-ring-builder object.builder create 5 3 108:18
yatinkota_: all change to current user and its 75508:18
kota_sounds good08:19
yatinkota_: all disk(vdi files) are of 2G, usage is 2%, doesn't look of storage08:19
yatinkota_: doens't look out of storage08:20
yatinkota_: this is my df -h line: /dev/sdb1       2.0G   33M  2.0G   2% /srv/node/sdb08:20
yatinkota_: /etc/fstab, /dev/sdb1/srv/node/sdb xfs noatime,nodiratime,nobarrier,logbufs=80 208:20
openstackgerritMahati Chamarthy proposed openstack/swift: Modify unit tests to include real crypto
kota_and "swift-ring-builder <builder file>" shows all disks you intend?08:21
yatinkota_: under /srv/node/sdb/, created 2 directories, sdb1 & sdb508:21
yatinkota_: using sdb1 into swift-ring-builder08:22
kota_just sdb1 only?08:22
yatinswift-ring-builder object.builder create 5 3 108:22
yatinswift-ring-builder object.builder add r1z1- 108:22
yatinkota_: yes, and other policy uses sdb508:22
yatinswift-ring-builder object-1.builder create 5 2 108:23
yatinswift-ring-builder object-1.builder add r1z1- 108:23
yatinkota_: oh! policy-0 used 4 disk08:23
kota_you should add more disks to ring08:23
yatinkota_: yes, default policy has 4 disk, as /sdb1 sdc2 sdd3 sde408:23
*** geaaru has joined #openstack-swift08:24
kota_that's good, each replica should be assigned to unique devices08:24
*** ppai has quit IRC08:24
yatinkota_: policy-0 ring08:24
yatinwift-ring-builder object.builder create 5 3 108:24
yatinswift-ring-builder object.builder add r1z1- 108:24
yatinswift-ring-builder object.builder add r1z2- 108:24
yatinswift-ring-builder object.builder add r1z3- 108:24
yatinswift-ring-builder object.builder add r1z4- 108:24
*** jordanP has joined #openstack-swift08:24
kota_and all (sdb1, sdc2, sdd3, sde4) are mounted, right?08:24
yatinkota_: yes, all mounted08:24
yatin/dev/sdb1       2.0G   33M  2.0G   2% /srv/node/sdb08:25
yatin/dev/sdc1       2.0G   33M  2.0G   2% /srv/node/sdc08:25
yatin/dev/sdd1       2.0G   33M  2.0G   2% /srv/node/sdd08:25
yatin/dev/sde1       2.0G   33M  2.0G   2% /srv/node/sde08:25
kota_ring shows sdb1 but you looks to mount it to sdb ?08:26
kota_looks like inconsistency between ring and actual mount point.08:26
yatinkota_: yes...this could be possible, let me know, what are all data you need from conf08:27
yatinkota_: i'm creating 2 directories into each disk, and giving that directories to ring08:27
kota_i think you should set devices from mount point path08:29
kota_like  swift-ring-builder object.builder add r1z1- 108:29
kota_and set "devices=/srv/node" in conf08:30
yatinkota_: I have four services for account, container, object and hence i have put account-sever/1.conf as /srv/node/sdb08:31
yatinkota_: /etc/swift/account-server/1.conf has similar 2.conf 3.conf 4.conf08:31
kota_if so, the mount points should be under /srv/node/sdb/08:32
kota_like /srv/node/sdb/sdb1 ...08:32
yatinkota_: want to use each service towards each disk, so, 1.conf says devices = /srv/node/sdb08:32
yatinkota_: 2.conf devices = /srv/node/sdc, 3.conf devices = /srv/node/sdd 4.conf devices = /srv/node/sde08:32
yatinkota_: not right?08:33
kota_i think your ring setting has enough capability for that08:34
yatinkota_: if i change to : swift-ring-builder object.builder add r1z1-
kota_each port has 1 device08:34
yatinkota_: that will give me 4 disk, I created directories to make it as 8 disk08:35
kota_that means each service has 1 device08:35
kota_ah, you mean you want to use 1 disk as 2 disks, right?08:35
yatinkota_: yes08:35
yatinkota_: by split into 2 dirs08:36
*** silor has joined #openstack-swift08:36
kota_i don't recommend such a use case (at least in production) but...08:36
*** venkat has quit IRC08:36
kota_if you want, you set disk as loop back08:37
yatinkota_: this is not prod, virtualbox want to extend saio example from 1 disk to 4 disks08:37
kota_false to mount check.08:37
yatinkota_: mount_check is false08:37
*** ppai has joined #openstack-swift08:37
kota_set your all devices (strictly directories you want) to ring.08:38
kota_i.e. you have to set 8 directories into your ring.08:38
kota_if split 4 devices to 8 devices.08:39
d0ugalI'm getting this error, which I thought was authentication related, but I am having no problems doing other operations. Is it possible there are restructions on doing a PUT? or what else can I check?
yatinkota_: this mount_check should be false all object too?08:41
kota_d0ugal: it doesn't look like authentication error.08:41
yatinkota_: i have it set true, could be problem?08:41
* yatin checking object put08:42
yatinkota_: worked...conf problem, mount_check false needed....:)08:43
d0ugalkota_: ah, I was mostly guessing based on something a colleague said. Do you know where I should try looking to debug it?08:43
*** silor has quit IRC08:43
kota_d0ugal: could you give me the response body? that would be helpful to know what was going on.08:44
kota_yatin: i think so08:44
kota_yatin: your conf setting suggests each disk should be mounted under /srv/node/sdb but you mounted under /srv/node/, right?08:45
* d0ugal hunts for it08:45
yatinkota_: no, i mounted under same path08:45
yatinkota_: please check df output above08:45
yatinkota_: node/ I just put node dir for convention08:46
kota_yatin: no no no08:47
kota_yatin: if you set "devices=/srv/node" swift assumes subdirs are mount points (i.e. /srv/node/sdb should be a mount point)08:48
*** venkat has joined #openstack-swift08:48
kota_yatin: but you set "devices=/srv/node/sdb". That means some mount points should be under /srv/node/sdb (i.e. likely /srv/node/sdb/sdb)08:49
yatinkota_: and same sdb given into ring-builder?08:49
d0ugalkota_: The full response body is just: <html><h1>Precondition Failed</h1><p>A precondition for this request was not met.</p></html>08:51
* d0ugal hunts for logs08:51
kota_yatin: sorry, i didn't get your mean08:52
kota_yatin: but if ring has /sdb and you made a dir /srv/node/sdb/sdb and you set devices=/srv/node/sdb08:53
kota_i think that work well.08:53
* yatin checking multi-node swift doc..brb08:53
kota_d0ugal: curious08:54
*** zhill has quit IRC08:54
yatinkota_: still not clear. devices=/srv/node/sdb having 2 dirs sdb1 and sdb5 and ring has sdb108:56
kota_yatin: you should add also sdb5 into your ring.08:56
yatinkota_: added for policy-108:56
yatinkota_: for object-1.builder08:57
yatinkota_: ring08:57
*** acoles_ is now known as acoles08:57
yatinkota_: 4 disk, have two dir each, 1 set of four dirs from 4 disk to default storage policy08:58
yatinkota_: and other set of 4 dirs from 4 different disk to another storage policy08:59
yatinkota_: want to test storage policy separation with different replications values...09:00
*** haomaiwang has quit IRC09:01
kota_all disks has subdirs?09:01
yatinkota_: correct09:01
mahaticacoles: hi, could you take a look at this sometime? it'll help move things forward.09:01
*** haomaiwang has joined #openstack-swift09:01
yatinkota_: thanks a ton for all the way your help...really appreciated09:01
*** Jeffrey4l has quit IRC09:01
yatinkota_: :)09:02
kota_yatin: no worries, sorry i couldn't complete09:02
yatinkota_:  next thing i want to try out if ACLs09:03
yatinkota_: -if09:03
yatinkota_: do you have a min? i got some quick questions09:04
*** Jeffrey4l has joined #openstack-swift09:04
yatinkota_: about tempAuth and account and container ACLs09:04
kota_d0ugal: that looks a default PreconditionFaild message....09:05
kota_yatin: yup09:05
d0ugalkota_: Yeah, I think so09:06
d0ugalkota_: I get it with the CLI when I try a second upload.
d0ugalso, it seems I can't update an object?09:06
yatinkota_: for pipeline tempAuth, swift defines account, user, password and role09:07
yatinkota_: in swift .admin .reseller_admin is called role or group?09:07
kota_i think "role" fits your mean.09:08
yatinkota_: so, if any user has no role assign, then to allow him read/write access swift provides ACLs. is that right?09:09
acolesgood morning09:09
mahaticgood morning09:09
kota_acoles: good morning09:09
kota_yatin: yes09:10
yatinacoles: good morning! its afternoon here ;)09:10
*** rohit_ has quit IRC09:10
yatinmahatic: good morning09:10
acolesnotmyname: (from scrollback) i have a bash alias amend='git commit -a --amend' :)09:11
kota_oh, yeah09:11
mahaticyatin: it's afternoon here too, I was wishing acoles ;)09:11
kota_revert my "but"09:11
yatinmahatic: good afternoon :)09:11
kota_yatin, d0ugal: sorry I'm running out of time.09:11
kota_shoud leave my office.09:12
acolesok so we need an international convention on timezone neutral greetings09:12
acolesyo! :)09:12
d0ugalkota_: np, thanks!09:12
yatinkota_: thanks for your time09:13
mahaticacoles: :D will remember to use that ;)09:13
acolesmahatic: ack re your patch, will try09:13
mahaticacoles: thanks09:13
yatinacoles: kudos!09:13
* acoles has a mismatch between todo list and available time :/09:14
kota_acoles: I noticed you wrote down your suggestion to change object listing, oh no. just I was making acks in inline.09:14
kota_to my container-listing patch.09:14
kota_acoles: anyways, i commit my commens, they are temtative tho. sorry.09:15
acoleskota_: which patch? what did i do wrong?09:15
* yatin going on swift acl tour to grab concepts and few questions09:15
kota_acoles: patch 19863409:15
patchbotkota_: - Support last modified on listing containers09:15
kota_acoles: not wrong. they are great comments but i didn't have enough time to consider them.09:16
kota_so sorry but much appreciated.09:16
kota_will see them again09:17
acoleskota_: oic. yeah, i just wasn't sure. actually, I tried to find a nicer way to implement the "ceil" in your func test, then gave up (what you have is fine!) and then noticed that the container HEAD is already rounded.09:17
acoleskota_: _sleep_enough - nice!09:24
*** jistr has joined #openstack-swift09:25
*** km has quit IRC09:28
openstackgerritHisashi Osanai proposed openstack/swift: WIP: Fix signal handling for daemons with InternalClient
*** openstackgerrit has quit IRC09:32
*** openstackgerrit has joined #openstack-swift09:33
ho^^ couldnt' reproduce the problem. i might misunderstand something...09:44
*** dmorita has joined #openstack-swift09:48
*** dmorita has quit IRC09:53
*** links has quit IRC09:54
openstackgerritAlistair Coles proposed openstack/swift: Sleep enough for trampoline
acoleskota_: ^^ just fixed typo in commit message then +209:58
*** haomaiwang has quit IRC10:01
*** haomaiwang has joined #openstack-swift10:01
*** esker has joined #openstack-swift10:01
hoacoles: you may fix the commit message from new gerrit (edit buttom then select "Commit Message")10:02
acolesho: yes, thats how i did it. but it does cause a new patchset version to be created. which makes sense i guess.10:03
acolesho: in fact, you could edit commit message in the old gerrit, now you can also edit patch files, which is scary imho10:04
hoacoles: i see. it's useful :-)10:04
acolesho: it concerns me that it encourages people to make changes to code on gerrit and therefore not run unit/pep8 checks locally before submitting the change.10:05
*** esker has quit IRC10:06
*** links has joined #openstack-swift10:06
*** nakagawamsa has quit IRC10:09
hoacoles: yeah, good ui has bad side effect...10:10
*** mahatic has quit IRC10:12
hoi will leave now. have a nice day & weekend!10:15
*** ho has quit IRC10:16
acolesho: thanks, and you10:20
*** trifon has quit IRC10:35
*** esker has joined #openstack-swift10:37
*** SkyRocknRoll has quit IRC10:39
*** esker has quit IRC10:42
d0ugalkota_: FYI, if you are curious, the error I had seemed to be related to versioned containers (and me probably using them wrongly)10:45
*** suresh has joined #openstack-swift10:46
sureshmattoliverau: Thnqq for your answer10:49
*** ekarlso has quit IRC10:51
*** ekarlso has joined #openstack-swift10:51
*** dmorita has joined #openstack-swift10:54
*** dmorita has quit IRC10:58
*** jordanP has quit IRC10:59
*** haomaiwang has quit IRC11:01
*** haomaiwang has joined #openstack-swift11:01
*** haomaiwang has quit IRC11:14
openstackgerritAlistair Coles proposed openstack/swift: Fix rst errors so that html docs are complete
acolesnotmyname: ^^ more disappearing doc11:32
*** yatin has quit IRC11:35
*** sanchitmalhotra has quit IRC11:36
*** suresh has quit IRC11:40
*** esker has joined #openstack-swift11:49
*** links has quit IRC11:50
*** silor has joined #openstack-swift11:50
*** esker has quit IRC11:54
*** links has joined #openstack-swift12:06
*** Jeffrey4l has quit IRC12:06
*** links has quit IRC12:16
openstackgerritBing Hu proposed openstack/python-swiftclient: Support uploading to an object in swift from stdin Closes-bug: #1521342
openstackbug 1521342 in python-swiftclient "support uploading from a stdin stream" [Wishlist,In progress] - Assigned to Bing Hu (hubingsh)12:22
*** esker has joined #openstack-swift12:25
*** esker has quit IRC12:30
*** venkat has quit IRC12:33
openstackgerritBing Hu proposed openstack/python-swiftclient: Support uploading to an object in swift from stdin Closes-bug: #1521342
openstackbug 1521342 in python-swiftclient "support uploading from a stdin stream" [Wishlist,In progress] - Assigned to Bing Hu (hubingsh)12:38
*** SkyRocknRoll has joined #openstack-swift12:38
*** venkat has joined #openstack-swift12:44
*** lyrrad has joined #openstack-swift12:45
*** ____bb has joined #openstack-swift12:49
*** ____bb has quit IRC12:52
*** fesp has joined #openstack-swift12:59
*** odsail has joined #openstack-swift13:00
*** odsail has quit IRC13:00
*** esker has joined #openstack-swift13:01
*** fesp has quit IRC13:03
*** esker has quit IRC13:06
*** jordanP has joined #openstack-swift13:15
*** links has joined #openstack-swift13:25
*** venkat has quit IRC13:25
*** thumpba has joined #openstack-swift13:32
*** breitz has quit IRC13:33
*** breitz has joined #openstack-swift13:34
*** esker has joined #openstack-swift13:37
*** esker has quit IRC13:41
*** haomaiwa_ has joined #openstack-swift13:46
*** links has quit IRC13:54
openstackgerritOpenStack Proposal Bot proposed openstack/swift: Updated from global requirements
*** ppai has quit IRC13:55
*** haomaiwa_ has quit IRC14:01
*** haomaiwang has joined #openstack-swift14:01
*** dustins|out is now known as dustins14:05
*** robefran has joined #openstack-swift14:07
*** Jeffrey4l has joined #openstack-swift14:11
*** esker has joined #openstack-swift14:13
*** esker has quit IRC14:18
*** yatin has joined #openstack-swift14:25
*** badari has joined #openstack-swift14:26
*** dslevin has quit IRC14:36
*** chsc has joined #openstack-swift14:43
*** trifon has joined #openstack-swift14:43
*** blmartin has joined #openstack-swift14:48
*** esker has joined #openstack-swift14:49
*** yatin has quit IRC14:52
*** esker has quit IRC14:53
*** yatin has joined #openstack-swift14:55
*** dmorita has joined #openstack-swift14:56
*** chsc has quit IRC14:57
*** haomaiwang has quit IRC15:01
*** dmorita has quit IRC15:01
*** haomaiwang has joined #openstack-swift15:01
*** chsc has joined #openstack-swift15:02
*** yatin has quit IRC15:04
*** lpabon has joined #openstack-swift15:05
*** david-lyle has quit IRC15:08
*** rledisez1 has joined #openstack-swift15:21
*** esker has joined #openstack-swift15:22
*** esker has quit IRC15:24
*** esker has joined #openstack-swift15:24
*** tsg has joined #openstack-swift15:34
*** chsc has quit IRC15:35
*** [1]eranrom has joined #openstack-swift15:36
*** eranrom has quit IRC15:39
*** [1]eranrom is now known as eranrom15:39
*** janonymous_ has joined #openstack-swift15:44
acolespeluse: might you get chance to look at patch 231121 one more time before holidays?15:45
patchbotacoles: - Make ECDiskFile report all fragments found on disk15:45
openstackgerritBill Huber proposed openstack/swift: Update Erasure Coding Overview doc to remove Beta version
*** SkyRocknRoll has quit IRC15:48
*** minwoob has joined #openstack-swift15:49
peluseacoles: yup, will do15:50
acolespeluse: thanks!15:50
gmmahakota_: thanks for the comment. Let me run the setup on a new box and confirm we dont need it.15:52
*** klrmn has joined #openstack-swift15:54
janonymous_Please review these patches: , ,
*** chinaer has joined #openstack-swift15:59
*** lcurtis has joined #openstack-swift15:59
*** haomaiwang has quit IRC16:01
*** haomaiwang has joined #openstack-swift16:01
*** blmartin has quit IRC16:06
peluseWTF, when did gerrit change?  I can't even find the button to copy a line to pull down a specific patch??16:07
*** blmartin has joined #openstack-swift16:07
janonymous_hahaa, on the right top corner under Download  i think.16:09
onovyyep, Download16:09
onovytop right16:09
peluseheh, thanks guys :)  I see it now16:09
onovybut it's terrible. for example reply button is on top of page? so you want to reply do last comment ... scroll to top to hit Reply -> PERFECT!16:10
acolesonovy: peluse: somebody needs to start a "give-me-back-the-gerrit-i-loved" petition :)16:12
peluseacoles: totally agree!  I don't think this new i/f.  it reminds me of windows 10 or something16:12
*** esker has quit IRC16:12
peluseyeah, the ultimate insult!16:13
acolespeluse: it reminds me of the kind of UI that a software engineer would design16:13
janonymous_Also there is no way to see dependent patches, i donno16:14
*** rcernin has quit IRC16:15
peluseyeah., what was wrong with the old UI anyway?  maaaaaan....16:15
acolespeluse: anyway, if its *my* patch you are looking at, hit keyboard shortcut 'a', +2, workflow +1 :P :P16:15
acolesjanonymous_: top right, related patches == dependent16:15
peluseacoles: I shall follow those directions precisely :)16:15
tdasilvajanonymous_: just a question on patch 208222, is there a way to catch 'warn' being used by developers, maybe during pep8 test or something?16:16
patchbottdasilva: - Python 3 deprecated the logger.warn method in favo...16:16
acolespeluse: that was totally a joke!!!16:16
peluseheh, no doubt acoles :)16:16
acolesc'mon, i'm not that brazen16:16
janonymous_acoles: Ohh.. :)16:16
janonymous_tdasilva: Updating the fake logger does this i guess.16:17
tdasilvajanonymous: but only for tests, no?16:18
janonymous_No, updating fake logger tells when warn is used in actual code also if the func is covered.16:19
onovywhy is this not done in pep8/hacking like assertEqual(s)?16:20
*** diazjf has joined #openstack-swift16:20
janonymous_onovy: there is no best way to do so, but it was probably one of the better options.16:22
janonymous_you could try using warn in code and run unittests if you want16:22
janonymous_it would blow up a warning to user saying to use warning16:23
onovybest way is to put it in hacking imho :)16:25
*** arnox has left #openstack-swift16:25
*** minwoob has quit IRC16:26
janonymous_onovy: this patch is already pending from 5 months now i guess :(16:27
*** zhill has joined #openstack-swift16:28
onovyjanonymous_, link? can't found it16:29
janonymous_onovy: just above,  tdasilva shared it.16:30
janonymous_patch 20822216:30
patchbotjanonymous_: - Python 3 deprecated the logger.warn method in favo...16:30
janonymous_tdasilva: can i help ?16:30
onovyehm. you don't understand me16:30
onovy // this hacking16:30
onovy // here it checks assertEquals16:31
tdasilvajanonymous: basically I had the same question as onovy, why not add to hacking?16:31
*** chinaer has quit IRC16:31
tdasilvacan it be done?16:31
onovyand my question is: why not add this same check for logger.warn16:31
onovyi think it can be done (simple)16:31
janonymous_okay, no problem then16:31
onovyhacking is for this checks :)16:32
janonymous_but i have to search a bit on how to add it16:33
onovynp. send me a review for hacking i will take a look16:34
onovy(when you will have it of course :)16:35
acolesnotmyname: reviews with a +2 and a -1 do not appear anywhere on the review dashboard, I think? e.g. patch 259165 ;)16:38
patchbotacoles: - remove beta label from EC16:38
janonymous_So, i would push up a change in hacking repo , but does that makes this patch dependent on that ?16:39
peluseWTF I commented -1 on patch 259165 and its not there anymore16:41
patchbotpeluse: - remove beta label from EC16:41
onovyjanonymous_, hacking can be patched and released very fast16:41
peluseoh wait, I commented on patch 24821016:41
patchbotpeluse: - Update Erasure Coding Overview doc to remove Beta ...16:41
*** petertr7_away is now known as petertr716:42
janonymous_tdasilva: please let me know what do u think..16:43
tdasilvajanonymous: i like the idea of adding to hacking16:45
janonymous_sure then, i would update hacking repo :)16:45
*** jistr has quit IRC16:49
*** petertr7 is now known as petertr7_away16:52
*** dmorita has joined #openstack-swift16:57
notmynamegood morning16:59
notmynameacoles: yeah, I need to revisit the dashboards after the gerrit upgrade17:00
janonymous_notmyname: o/17:00
*** haomaiwang has quit IRC17:01
*** nadeem has joined #openstack-swift17:01
*** dmorita has quit IRC17:01
*** haomaiwang has joined #openstack-swift17:01
acolesnotmyname: my dashboard broke :/ i was missing some label: prefixes before Workflow and Verified.17:01
*** jordanP has quit IRC17:03
*** klrmn has quit IRC17:03
*** zhill has quit IRC17:05
*** gyee has joined #openstack-swift17:06
*** yatin has joined #openstack-swift17:09
*** robefran has quit IRC17:15
*** nadeem has quit IRC17:16
*** nadeem has joined #openstack-swift17:17
openstackgerritBill Huber proposed openstack/swift: Update Erasure Coding Overview doc to remove Beta version
notmynameacoles: thanks for the doc updates. approved17:20
acolesnotmyname: is weird, build_sphinx spews out errors but then reports success, so gate passes17:21
notmynamewbhuber: re patch 24821017:24
patchbotnotmyname: - Update Erasure Coding Overview doc to remove Beta ...17:24
notmynamewbhuber: I agree with what acoles said on patch 25916517:24
patchbotnotmyname: - remove beta label from EC17:24
*** rohit_ has joined #openstack-swift17:24
notmynamewbhuber: don't try to invent something between beta and prod, and don't make forward-looking statements. just leave it as "you can use it"17:25
notmynamewbhuber: I definitely think yours is better than mine because (1) yours came first and (2) yours has the change to the ssync docs17:25
notmynamewbhuber: what would you think about dropping more of the words at the top of the EC doc?17:26
openstackgerritMerged openstack/swift: Not used parameter
wbhubernotmyname: you got it.17:27
wbhubernotmyname: will remove the entire section at the top of the EC doc now and resubmit a patch in a few.17:27
notmynamewbhuber: hey, that's, just, like, my opioning, man17:28
wbhubernotmyname: i think that makes more sense.  your, peluse's and alistar's opinions are more than just mine.  gotta go with the majority, right?17:29
notmynamewbhuber: I abandoned my patch17:30
wbhubernotmyname: nifty.17:30
notmynamewbhuber: but another patch set on yours, and I'll +217:30
openstackgerritBill Huber proposed openstack/swift: Update Erasure Coding Overview doc to remove Beta version
notmynamewbhuber: boom. there it is :-)17:32
notmynameacoles: could you take 60 seconds and look at that one too, please?17:33
*** zhill has joined #openstack-swift17:34
acoleswait, this is whuber's?17:35
wbhuberyes, wbhuber.17:36
wbhuberps:// -17:36
claygnotmyname: I'm sure acoles thinks his fixes for EC should go in before we take off the beta label17:36
*** trifon has quit IRC17:37
claygacoles: well not sure, you care about the label - but sure you want those patches reviewed ;)17:37
*** zhill has quit IRC17:38
acoleswbhuber: got it, thanks17:40
acolesclayg: sure I do :)17:40
acolesclayg: but i thought about it and i don't think there is anything *broken* that they fix, just performance improvements17:41
acolesclayg: (although its been a while since i worked on them)17:41
acolesclayg: actually, i stumbled across a nice corner case while sorting out the mismatched suffix hashing that you spotted with the fast-post patch17:42
openstackgerritBill Huber proposed openstack/swift: Update Erasure Coding Overview doc to remove Beta version
acolesi fixed the hashing to deal with the differing meta filename combinations, added some assertions to probe tests, and... boom...17:43
wbhuberacoles: i like your object-server.conf sample recommendation.  made a single change to the newest patch.17:43
wbhubernotmyname: that means you're needed to check it out once again.17:44
acolesclayg: ssync will not reconcile a node with (t2.meta, t0.ts) and another with just (t0.ts) ... and I *think* that could happen with failing nodes, handoffs taking PUTs etc17:45
*** yatin has quit IRC17:46
*** diogogmt has joined #openstack-swift17:47
*** yatin has joined #openstack-swift17:48
peterlisakhello, is someone available to review two small patches: and
*** trifon has joined #openstack-swift17:53
*** david-lyle has joined #openstack-swift17:54
*** zhill has joined #openstack-swift17:54
*** rledisez has quit IRC17:55
*** robefran has joined #openstack-swift17:56
*** david-lyle has quit IRC17:59
*** david-lyle has joined #openstack-swift18:00
*** haomaiwang has quit IRC18:01
*** haomaiwang has joined #openstack-swift18:01
*** diogogmt has quit IRC18:02
*** diogogmt has joined #openstack-swift18:03
*** rcernin has joined #openstack-swift18:05
openstackgerritMerged openstack/swift: Fix rst errors so that html docs are complete
*** klrmn has joined #openstack-swift18:11
*** yatin has quit IRC18:19
*** dmorita has joined #openstack-swift18:24
*** dmorita has quit IRC18:25
*** diogogmt has quit IRC18:25
*** dmorita_ has joined #openstack-swift18:25
*** zaitcev has joined #openstack-swift18:26
*** ChanServ sets mode: +v zaitcev18:26
notmynameclayg: peluse: I'd love your opinion (in gerrit, as a review comment) on patch 248210 :-)18:29
patchbotnotmyname: - Update Erasure Coding Overview doc to remove Beta ...18:29
pelusenotmyname: OK, on a call but will take a look when I'm done18:29
*** HenryG is now known as HenryG_afk18:29
*** rcernin has quit IRC18:36
*** janonymous_ has quit IRC18:37
*** geaaru has quit IRC18:38
*** diogogmt has joined #openstack-swift18:41
*** aix has quit IRC18:56
*** haomaiwang has quit IRC19:01
*** haomaiwang has joined #openstack-swift19:01
acolesgood night19:03
*** diazjf has quit IRC19:04
*** acoles is now known as acoles_19:06
*** nadeem has quit IRC19:07
flaper87notmyname: (et al)
*** openstackgerrit has quit IRC19:17
*** openstackgerrit has joined #openstack-swift19:18
*** nadeem has joined #openstack-swift19:24
*** zaitcev has quit IRC19:30
*** nadeem has quit IRC19:35
*** zaitcev has joined #openstack-swift19:37
*** ChanServ sets mode: +v zaitcev19:37
*** aix has joined #openstack-swift19:38
*** HenryG_afk is now known as HenryG19:42
*** diazjf has joined #openstack-swift19:49
*** zaitcev has quit IRC19:55
*** haomaiwang has quit IRC20:01
*** haomaiwang has joined #openstack-swift20:01
*** esker has joined #openstack-swift20:01
*** esker has joined #openstack-swift20:02
*** petertr7_away is now known as petertr720:02
*** esker has quit IRC20:02
*** esker has joined #openstack-swift20:03
*** zaitcev has joined #openstack-swift20:09
*** ChanServ sets mode: +v zaitcev20:09
*** torgomatic has joined #openstack-swift20:15
*** ChanServ sets mode: +v torgomatic20:15
notmynameZyric: hello20:15
Zyricnotmyname: I had another look at patch 241978 and changing the primary name of a policy with only one name to something new doesn't appear to remove the old name or cause errors20:17
patchbotZyric: - Add support for storage policies to have more than...20:17
ZyricBecause names are only removed if the new name isn't equal to (which is the same as the first element) and also is in the .alias_list20:19
notmynameZyric: I'll have to look again (and this time run some tests)20:19
*** jamielennox is now known as jamielennox|away20:21
*** lyrrad has quit IRC20:21
*** esker has quit IRC20:22
*** zaitcev has quit IRC20:25
*** robefran has quit IRC20:25
*** zaitcev has joined #openstack-swift20:36
*** ChanServ sets mode: +v zaitcev20:36
notmynamewe made an infographic
*** zhill has quit IRC20:40
onovynotmyname, nice!20:44
onovywhat metric did you choose for stackalytics graph? loc, commits, ...?20:44
notmynameonovy: IIRC it's
onovyah, effort. thanks20:45
notmynamethe person-day effort is the only one there that might be good to actually track. I don't want to encourage anyone to get commits or comments on lines of code just to go up in that ranking!20:46
onovyof course20:46
*** dmorita_ has quit IRC20:47
pdardeaunotmyname: very nice!20:47 14th :), but in Mitaka only: 4th place20:47
*** petertr7 is now known as petertr7_away20:48
notmynameonovy: :-)20:48
pdardeaunotmyname: either contributors or contributers is spelled incorrectly. not sure which. i used to be ok at speling ;-)20:50
claygnot sure how to submit patches to blog images :\20:51
notmynameI do. new versions should be up momentarily ;-)20:52
notmynameprocess involves me walking over to the design person in marketing. very manual. no scripting or automation at all20:52
*** dmorita has joined #openstack-swift20:55
pdardeaunotmyname: lol. sweet!20:55
notmynamepdardeau: it's fixed now. hit reload20:55
*** petertr7_away is now known as petertr720:56
*** zhill has joined #openstack-swift20:57
pdardeaunotmyname: i saw it! very nice. swank! ;)20:57
*** robefran has joined #openstack-swift20:58
*** haomaiwang has quit IRC21:01
*** silor has quit IRC21:01
*** gyee has quit IRC21:01
*** 17WABC36L has joined #openstack-swift21:01
onovy do we need reply from some else? swauth is ok.21:02
notmynameonovy: thanks21:04
blmartinOh, onovy, I answered your question about the force option on the review page21:04
onovyblmartin, yep. you are right, my arguments are invalid :)21:05
blmartinwell not invalid so much as a different take on the matter. I think it is valid and something that may or may not be wanted. I'm not really sure21:06
onovynotmyname, exactly i didn't fix it in swauth. gholt already did it few months/years ago :)21:06
onovyblmartin, your solution is compatible with original approach. so go this way and maybe change it later in another commit?21:07
blmartinThat sounds fair to me21:07
*** blmartin has left #openstack-swift21:18
*** dmorita has quit IRC21:24
*** changbl has quit IRC21:26
*** nadeem has joined #openstack-swift21:30
*** nadeem has quit IRC21:31
*** nadeem has joined #openstack-swift21:31
*** gyee has joined #openstack-swift21:32
*** diogogmt has quit IRC21:32
*** diogogmt has joined #openstack-swift21:33
*** changbl has joined #openstack-swift21:39
*** petertr7 is now known as petertr7_away21:46
*** dmorita has joined #openstack-swift21:49
notmynameonovy: still here?21:49
notmynameI think I found what I was looking for. patch 22953221:50
patchbotnotmyname: - swift-init: SIGKILL daemon after kill_wait period.21:51
onovybut 10:50 pm here :)21:51
onovyperfect, what you found?21:51
notmynameonovy: I was thinking about the change a little more, and I don't want to change the current behavior, I don't think. ie the really-kill thing is good. but not with the same command as the existing one21:52
*** trifon has quit IRC21:52
notmynameso I was bringing it back up to rememeber which way you had implemented it21:52
notmynameand to leave that comment if it was implemented as a kill-after-timeout on the existing command21:52
onovyso add option --kill-after-timeout ?21:53
notmynameyeah, something like that21:53
onovyi'm ok with this option. i will use it inside debian init scripts and everything will be perfect21:53
notmynameright. and that way it won't impact existing tools and workflows, and it also avoids questions about graceful restarts21:54
onovyperfect, will you write it to gerrit?21:55
notmynameonovy: just brainstorming here, what about a `swift-init <server> stop --kill-after-timeout=30` version? then swift-init passes that timeout to a manager, very similarly to how you're doing it now21:55
notmynameI don't know if it's good to expose that timeout or not21:55
onovytimeout can be changed now21:55
onovy--timeout param i think21:56
notmynameah ok21:56
onovytwo timeouts are useless :)21:56
onovythis option will work with: stop, restart, reload, gracefull-stop, right?21:56
onovyand in all commands will kill process __group__ after timeout?21:56
onovygracefull-stop = shutdown21:57
notmynamemaybe not reload. that's just supposed to reload the configs, right21:57
onovyreload = shutdown + start21:57
notmyname(It's been a while since I looked at this part. I could be forgetting something)21:57
onovyreload = stop (gracefull) + start21:57
notmynameah, right. reload = graceful stop + start, right?21:57
notmynameand restart = kill + start21:57
onovyshutdown = stop(gracefull)21:58
claygnotmyname: well sigterm21:58
notmynameok, thanks. I'll leave a comment in gerrit21:58
claygshutdown is HUP21:58
onovystop=sigterm, stop(gracefull)=sighup21:58
*** 17WABC36L has quit IRC22:01
notmynamecomment left22:01
*** haomaiwang has joined #openstack-swift22:01
openstackgerritMerged openstack/swift: Update Erasure Coding Overview doc to remove Beta version
onovynotmyname, thank you!22:03
openstackgerritMerged openstack/swift: Deprecated param timeout removed from memcached
openstackgerritMerged openstack/swift: Fix incorrect kwarg in auth middleware example
notmynamewhoa. flurry of merges22:05
*** gyee has quit IRC22:06
*** petertr7_away is now known as petertr722:10
*** petertr7 is now known as petertr7_away22:21
*** badari has quit IRC22:28
notmyname#success removed the beta tag on the swift erasure code docs22:33
openstackstatusnotmyname: Added success to Success page22:33
*** nadeem has quit IRC22:36
*** rcernin has joined #openstack-swift22:52
*** lcurtis has quit IRC22:54
openstackgerritOndřej Nový proposed openstack/swift: swift-init: New option kill-after-timeout
onovyand now i'm going to bed :)22:58
*** lpabon has quit IRC22:58
*** haomaiwang has quit IRC23:01
*** haomaiwang has joined #openstack-swift23:01
*** trifon has joined #openstack-swift23:01
*** rcernin has quit IRC23:08
*** rohit_ has quit IRC23:10
*** ianbrown has quit IRC23:12
*** ianbrown has joined #openstack-swift23:12
*** diazjf has quit IRC23:13
*** SorinSen has quit IRC23:17
*** mac_ified has joined #openstack-swift23:18
then3rdGood afternoon, all.23:18
notmynameZyric: yup, I agree and have understand better on patch 24197823:25
patchbotnotmyname: - Add support for storage policies to have more than...23:25
then3rdHowdy. I have another general question I've been thinking about.23:25
Zyricnotmyname: Neat, thank you!23:25
then3rdIf I want to design a lean [hardware] cluster, is it common to combine roles/tiers? Here's sort of what I have in mind. Two types of proxy nodes. First, a combination haproxy/swift-proxy node. Simply add more of these if a region needs to handle more traffic; haproxy distributes load to backend swift proxies via the 10G network. The other type of proxy node would still run haproxy/swift in addition23:27
then3rdto keystone, nova, horizon and supporting services. Currently I can't imagine needing more than three or four proxies total. I don't want to "waste" hardware by having standalone controller nodes. 2+ required. Storage nodes would only need to be powerful enough to handle the storagere/replications services.23:27
notmynamethen3rd: right23:28
notmynamethen3rd: there's generally 3 different SKUs you can end up with in a swift cluster23:28
notmynameI know I've written this down before. let me see if I can find it23:28
then3rdOh, excellent.23:28
notmynamethen3rd: this should help
*** nadeem has joined #openstack-swift23:30
then3rdSince I didn't do the initial architechture, I haven't considered hardware until now.23:30
then3rdThat was a helpful article. I have a feeling some of my hardware isn't enough and some is overkill.23:40
notmynamethen3rd: swift is pretty flexible. note that the swiftstack docs will push to opinionated recommendations, both to ensure success and to keep our support burden lower23:41
notmynamethen3rd: you can see that what you have now is working. and that's great!23:42
*** nadeem has quit IRC23:42
then3rdIs it unusual to run haproxy on the same nodes as swift proxy? The vip would only be active on the primary, and balacnes between localhost and the other swift proxies via the cluster network.23:43
then3rdBecause that's how it was built. :D23:44
notmynameno, that's fine23:44
then3rdAs long as I don't run out of bandwidth/cpu/memory. XD23:44
* clayg considers the set of things possible when removing the constraint of running out of bandwidth/cpu/memory23:47
*** dmorita has quit IRC23:47
notmynameclayg: "everything", I think23:48
then3rdheh. How large are the clusters some of you guys operate?23:50
notmyname(technically, it's the customers who are operating them with tools we write) all sizes. everything from a few dozen TB to many PB23:52
*** dmorita has joined #openstack-swift23:55
Zyricnotmyname: I have an account watcher mostly working now, just finishing tests - but there's a lot of duplicate code that probably could be shared with the object watcher.23:56
ZyricShould I attempt to unify the two or leave them separate for now, maybe leave that for another patch?23:56
notmynameZyric: I'd suggest leaving it separate for now, but expect that the patch that lands will have them unified23:56
notmynameZyric: ie expect torgomatic's patch to land and you'll be able to use what's in there (or refactor to a common place)23:57
notmynameZyric: point is, don't get distracted by the perfect design instead of making progress towards solving the problem23:57
Zyricnotmyname: Cool, that's similar to what I was thinking.23:58
ZyricAt the moment all it does is collect the account paths, is there better/other data that would be good to collect?23:59

Generated by 2.14.0 by Marius Gedminas - find it at!