Tuesday, 2016-01-05

*** haomaiwang has quit IRC00:01
*** haomaiwang has joined #openstack-swift00:01
notmynameonovy: yeah, I think that looks good. much better option with this patch set. thanks00:06
*** Vinsh has quit IRC00:23
*** Vinsh has joined #openstack-swift00:24
*** zhill has joined #openstack-swift00:31
*** Jeffrey4l__ has quit IRC00:39
*** rminmin has joined #openstack-swift00:46
*** diogogmt has joined #openstack-swift00:56
*** haomaiwang has quit IRC01:01
*** haomaiwang has joined #openstack-swift01:01
*** nakagawamsa has joined #openstack-swift01:06
*** rminmin has quit IRC01:09
mattoliveraukota_: o/01:13
kota_mattoliverau: o/01:13
*** garthb has quit IRC01:27
*** km_ has joined #openstack-swift01:28
*** km__ has quit IRC01:29
*** badari has quit IRC01:32
*** Vinsh has quit IRC01:32
*** Vinsh has joined #openstack-swift01:33
*** zhill has quit IRC01:38
*** haomaiwang has quit IRC02:01
*** haomaiwang has joined #openstack-swift02:01
*** diazjf has joined #openstack-swift02:03
*** doxavore has joined #openstack-swift02:04
kota_nakagawamasa: still?02:12
*** klrmn has quit IRC02:12
kota_nakagawamasa: https://bitbucket.org/kmgreen2/pyeclib/commits/4bff4a92ea2ab336081737f9f1acc1a3c949af78 might save you quickly on patch 20779002:13
patchbotkota_: https://review.openstack.org/#/c/207790/ - Update SAIO setup for Ubuntu 14.0402:13
kota_nakagawamasa: newest pyeclib doesn't include liberasurecode tar ball so an user should install via OS packaging system like apt or yum.02:15
*** zhill has joined #openstack-swift02:29
*** zhill has quit IRC02:32
*** zhill has joined #openstack-swift02:32
*** zhill has quit IRC02:36
*** badari has joined #openstack-swift02:42
*** Vinsh_ has joined #openstack-swift02:46
nakagawamsakota_: thanks infomation! I'll retry unittest with "sudo apt-get install liberasurecode-dev"02:48
*** Vinsh has quit IRC02:50
*** haomaiwang has quit IRC03:01
*** haomaiwang has joined #openstack-swift03:01
*** badari has quit IRC03:01
*** diazjf has quit IRC03:05
openstackgerritJanie Richling proposed openstack/swift: Fix POST-as-COPY mess up the DLO manifest  https://review.openstack.org/25279103:11
*** Jeffrey4l has joined #openstack-swift03:11
jrichli^^ rebase only03:11
*** diazjf has joined #openstack-swift03:12
*** diazjf has quit IRC03:13
*** klrmn has joined #openstack-swift03:19
openstackgerrityapeng Yang proposed openstack/swift: Replace assertEqual(None, *) with assertIsNone  https://review.openstack.org/26354703:24
*** Jeffrey4l has quit IRC03:26
*** km_ has quit IRC03:26
*** robefran has quit IRC03:26
*** km_ has joined #openstack-swift03:27
*** Jeffrey4l has joined #openstack-swift03:28
*** sanchitmalhotra has joined #openstack-swift03:28
*** Jeffrey4l has quit IRC03:42
*** Jeffrey4l has joined #openstack-swift03:43
*** doxavore has quit IRC03:44
*** venkat has joined #openstack-swift03:46
*** links has joined #openstack-swift03:52
*** haomaiwang has quit IRC04:01
*** haomaiwang has joined #openstack-swift04:01
*** trifon has joined #openstack-swift04:05
*** haomaiwang has quit IRC04:13
*** Jeffrey4l has quit IRC04:21
*** ppai has joined #openstack-swift04:24
*** suresh_ has joined #openstack-swift04:34
*** Jeffrey4l has joined #openstack-swift04:34
suresh_I am getting warning NOTE: Dispersion of 64.062500 indicates some parts are not optimally dispersed.04:35
suresh_while doing swift-ring-builder account.builder rebalnce04:36
suresh_please some one help04:36
*** trifon has quit IRC04:39
*** tsg_ has quit IRC04:42
*** zhill has joined #openstack-swift04:48
*** haomaiwang has joined #openstack-swift04:53
*** SkyRocknRoll has joined #openstack-swift04:54
openstackgerrityapeng Yang proposed openstack/swift: Replace assertEqual(None, *) with assertIsNone  https://review.openstack.org/26354704:56
*** haomaiwang has quit IRC05:01
*** haomaiwang has joined #openstack-swift05:01
*** PsionTheory has quit IRC05:03
*** miurahr has joined #openstack-swift05:11
*** zhill has quit IRC05:13
*** yatin has joined #openstack-swift05:27
*** ppai has quit IRC05:29
openstackgerrityapeng Yang proposed openstack/swift: Use assertTrue/False instead of assertEqual(T/F)  https://review.openstack.org/26356005:32
*** ppai has joined #openstack-swift05:42
*** klrmn has quit IRC05:43
*** yarkot has joined #openstack-swift05:55
*** km_ has quit IRC05:55
*** km_ has joined #openstack-swift05:57
*** haomaiwang has quit IRC06:01
*** haomaiwang has joined #openstack-swift06:01
*** gyee has quit IRC06:01
*** trifon has joined #openstack-swift06:05
*** joker_ has quit IRC06:06
*** peterlisak has joined #openstack-swift06:11
*** lcurtis has quit IRC06:11
openstackgerrityapeng Yang proposed openstack/swift: Replace assertEqual(None, *) with assertIsNone  https://review.openstack.org/26354706:13
*** joker_ has joined #openstack-swift06:18
*** venkat has quit IRC06:19
*** yarkot has quit IRC06:20
openstackgerrityapeng Yang proposed openstack/swift: Use assertTrue/False instead of assertEqual(T/F)  https://review.openstack.org/26356006:22
*** blmartin has joined #openstack-swift06:25
*** venkat has joined #openstack-swift06:31
*** Jeffrey4l has quit IRC06:36
suresh_I am getting warning NOTE: Dispersion of 64.062500 indicates some parts are not optimally dispersed. while doing swift-ring-builder account.builder rebalnce06:36
suresh_please some one help06:37
*** Jeffrey4l has joined #openstack-swift06:37
*** ChubYann has quit IRC06:44
kota_suresh_: docs are here, http://docs.openstack.org/developer/swift/overview_ring.html#dispersion06:45
kota_suresh_: for that case, it would be good to review device weight in your ring or overload value.06:46
kota_suresh_: docs about overload is a bit above in the dispersion document.06:47
*** blmartin has quit IRC06:55
*** km_ has quit IRC06:58
*** km_ has joined #openstack-swift06:59
*** haomaiwang has quit IRC07:01
*** 18WABPMMU has joined #openstack-swift07:01
openstackgerrityapeng Yang proposed openstack/swift: Use assertTrue/False instead of assertEqual(T/F)  https://review.openstack.org/26356007:22
*** Jeffrey4l has quit IRC07:30
*** venkat has quit IRC07:40
*** zaitcev has quit IRC07:42
*** Jeffrey4l has joined #openstack-swift07:51
*** venkat has joined #openstack-swift07:52
*** 18WABPMMU has quit IRC08:01
*** haomaiwang has joined #openstack-swift08:01
*** arnox has joined #openstack-swift08:03
*** eranrom has joined #openstack-swift08:08
*** rledisez has joined #openstack-swift08:12
*** Jeffrey4l has quit IRC08:18
*** Jeffrey4l has joined #openstack-swift08:21
*** eranrom has quit IRC08:30
*** eranrom has joined #openstack-swift08:39
*** m_kazuhiro has joined #openstack-swift08:45
*** m_kazuhiro has quit IRC08:46
*** haomaiwang has quit IRC08:53
*** haomaiwang has joined #openstack-swift08:53
*** jsuchome has joined #openstack-swift08:59
*** haomaiwang has quit IRC09:01
*** haomaiwang has joined #openstack-swift09:01
*** jordanP has joined #openstack-swift09:05
openstackgerritMerged openstack/swift: Port swift.common.utils.StatsdClient to Python 3  https://review.openstack.org/23700509:14
openstackgerritMerged openstack/swift: Fix AttributeError for LogAdapter  https://review.openstack.org/26161909:15
*** jistr has joined #openstack-swift09:21
*** km_ has quit IRC09:21
*** janonymous has joined #openstack-swift09:23
*** joeljwright has joined #openstack-swift09:29
*** ChanServ sets mode: +v joeljwright09:29
eranromcschwede: Have you abondoned the notification imddleware? Going over the comments seems like it got stuck over the problems raised with the need to be reliable form the Swift side. Any news?09:31
cschwedeeranrom: it’s not (yet) abandoned: https://review.openstack.org/#/c/196755/ however, i’m currently not working on it. reliability is one challenge (though doable, imo), but there are other missing parts. have a look at the last two comments from Sam in the review09:37
openstackgerritKota Tsuyuzaki proposed openstack/swift: Last-Modified header support on HEAD/GET container  https://review.openstack.org/19863309:42
*** silor has joined #openstack-swift09:44
eranromcschwede: thanks. Yes, as far as I understand one remark has to do with the reliability problem - which if I get right it says keep all in memory and forget about being 100% relaible. the other one seems specific for queues that enforce authorization. Which IMHO cover only part of the use case.09:45
cschwedeeranrom: well, to work with Zaqar authentication is required IIRC, so this is a hard problem. Just sending out notifications to a MQ (instead of sending it to Zaqar) is not that useful for public providers, though it might be enough for a private cloud with specific requirements. but then the question raised which MQ should be supported?09:52
*** jmccarthy has quit IRC09:54
*** silor1 has joined #openstack-swift09:54
*** jmccarthy has joined #openstack-swift09:55
*** silor has quit IRC09:55
*** silor1 is now known as silor09:55
jsuchomekota_: hi, any reason for keeping this waiting? https://review.openstack.org/#/c/189258/09:58
*** haomaiwang has quit IRC10:01
*** haomaiwang has joined #openstack-swift10:01
eranromcschwede,: Anything wrong with the ability to just configure a URL for POSTs? Do you think all the problems must be solved before this can land? I can dig deeper into IBM's use case, but seems to me there is a lot of value here even without solving all of the problems.10:18
openstackgerritOpenStack Proposal Bot proposed openstack/swift: Updated from global requirements  https://review.openstack.org/8873610:31
openstackgerritOndřej Nový proposed openstack/swift: swift-init: New option kill-after-timeout  https://review.openstack.org/22953210:34
*** hseipp has joined #openstack-swift10:37
*** hseipp has quit IRC10:37
*** hseipp has joined #openstack-swift10:37
kota_jsuchome: sorry just dropped from my heads. Will see tomorrow.10:40
kota_jsuchome: thx, for bringing it up to me.10:41
cschwedeeranrom: well, we need an agreement what the middleware should be doing. just sending a POST (without a valid Keystone token) is not enough for Zaqar, thus making the middleware useless except when you build your own receiving consumer. and this looks to specialized to me - if we ship a notification middleware with swift itself, it should be usable with some existing broker10:57
cschwedeeranrom: however, we could modify the middleware to use an own account to send notifications to zaqar (ie, the middleware needs to authenticate itself to keystone, fetch a token and use that token instead of forwarding a user token)10:58
cschwedeeranrom: and if there is no user/password defined for the middleware, no token is sent. that might be a suitable solution?+10:58
*** haomaiwang has quit IRC11:01
*** haomaiwang has joined #openstack-swift11:01
*** aix has quit IRC11:25
*** acoles_ is now known as acoles11:34
*** daemontool has joined #openstack-swift11:35
openstackgerritTomáš Matlocha proposed openstack/swift: Additional error log message for drive-audit  https://review.openstack.org/26369111:37
*** Jeffrey4l has quit IRC11:48
*** haomaiwang has quit IRC12:01
*** haomaiwang has joined #openstack-swift12:01
*** yatin has quit IRC12:11
*** nakagawamsa has quit IRC12:55
*** haomaiwang has quit IRC13:01
*** haomaiwang has joined #openstack-swift13:01
*** ppai has quit IRC13:02
*** kei_yama has quit IRC13:04
*** links has quit IRC13:10
*** haomaiwang has quit IRC13:12
*** aix has joined #openstack-swift13:14
*** proteusguy has quit IRC13:17
joker_ı am working on this bug https://bugs.launchpad.net/swift/+bug/150942913:21
openstackLaunchpad bug 1509429 in OpenStack Object Storage (swift) "X-Timestamp missing from object PUT/COPY response headers" [Low,Confirmed] - Assigned to Mustafa (celikmustafa89)13:21
joker_ı complete x-timestamp headers for account and container operations13:21
*** yatin has joined #openstack-swift13:22
joker_but for object operation  ı have problem with POST and PUT operations13:22
joker_if someone has experience about this i am gone give the details13:23
*** venkat has quit IRC13:25
*** proteusguy has joined #openstack-swift13:29
*** yatin has quit IRC13:36
*** haomaiwang has joined #openstack-swift13:40
*** haomaiwang has quit IRC13:45
*** dslevin has quit IRC13:45
*** suresh_ has quit IRC13:49
acoleseranrom: fyi i am mid way through reviewing patch 205803 so you may want to wait before addressing mattoliverau's review comments13:55
patchbotacoles: https://review.openstack.org/#/c/205803/ - Container-Sync to iterate only over synced containers13:55
*** SkyRocknRoll has quit IRC13:57
*** robefran has joined #openstack-swift14:01
*** dslevin_ has quit IRC14:02
*** 18VAALI1N has joined #openstack-swift14:07
openstackgerritLiuNanke proposed openstack/swift: Replace assertTrue(isinstance()) with assertIsInstance()  https://review.openstack.org/26374214:08
*** SkyRocknRoll has joined #openstack-swift14:08
*** petertr7_away is now known as petertr714:13
*** pdion891 has joined #openstack-swift14:21
*** links has joined #openstack-swift14:24
*** ccontini has joined #openstack-swift14:26
*** ccontini has quit IRC14:26
openstackgerritLiuNanke proposed openstack/swift: Replace assertTrue(isinstance()) with assertIsInstance()  https://review.openstack.org/26374214:26
*** ccontini has joined #openstack-swift14:28
*** tsg has joined #openstack-swift14:29
*** Vinsh_ has quit IRC14:30
*** Vinsh has joined #openstack-swift14:30
*** badari has joined #openstack-swift14:31
ccontiniHi all, I have a question about the replication in swift (2.2.0)14:34
ccontiniWe have a ring built like this14:35
ccontini33554432 partitions, 3.000000 replicas, 1 regions, 4 zones, 139 devices, 0.00 balance14:35
ccontiniThe minimum number of hours before a partition can be reassigned is 114:35
*** Vinsh_ has joined #openstack-swift14:36
*** Vinsh has quit IRC14:36
ccontiniOne of the object servers has been having a lot of issues and had a lot of drives changed + was rebooted several times14:36
ccontiniNow, I see that in the logs: Jan  5 09:32:31 swift04 object-replicator: 86955/6597974 (1.32%) partitions replicated in 65319.46s (1.33/sec, 1358h remaining)14:37
ccontiniIt looks like replication is taking forever14:37
ccontiniIs there any way we can speed up the process ?14:37
ccontiniI know the min_part_hours in the ring should be increased (a lot in our case) but I don't understand exactly what it does14:38
*** diogogmt has quit IRC14:43
ahalethe min part stuff is really just to protect against doing crazy stuff, to avoid having the ring move more than one copy of a partition during successive rebalances. i think the idea is to set it around your replication pass time14:47
ahalethat seems like quite a high part power for that device count, maybe theres a ton of slow directory walking going on.. i don't know - im not great at making replication go fast14:48
ccontiniBut from the log it looks like replication takes over 1358h right ?14:49
*** links has quit IRC14:49
ccontiniThat looks incredibly slow14:49
*** Jeffrey4l has joined #openstack-swift14:54
*** lcurtis has joined #openstack-swift14:57
*** blmartin has joined #openstack-swift14:58
*** 18VAALI1N has quit IRC15:01
*** haomaiwang has joined #openstack-swift15:01
*** Vinsh_ is now known as Vinsh15:04
*** breitz has quit IRC15:12
*** breitz has joined #openstack-swift15:12
*** jmccarthy1 has joined #openstack-swift15:13
*** jmccarthy1 has quit IRC15:13
*** jmccarthy1 has joined #openstack-swift15:13
*** jmccarthy1 has quit IRC15:14
*** jmccarthy1 has joined #openstack-swift15:15
*** barker has joined #openstack-swift15:16
*** jmccarthy has quit IRC15:16
*** Jeffrey4l has quit IRC15:19
*** SkyRocknRoll has quit IRC15:19
*** changbl has quit IRC15:27
openstackgerritMahati Chamarthy proposed openstack/swift: Modify unit tests to include real crypto  https://review.openstack.org/21145115:28
*** barker_ has joined #openstack-swift15:28
*** barker has quit IRC15:29
*** breitz has quit IRC15:29
*** breitz has joined #openstack-swift15:30
*** Jeffrey4l has joined #openstack-swift15:31
*** barker_ has quit IRC15:33
mahaticacoles: jrichli ^15:37
acolesmahatic: ack15:37
jrichlimahatic: great, I will take a look15:37
*** SkyRocknRoll has joined #openstack-swift15:40
*** Jeffrey4l has quit IRC15:41
*** SkyRocknRoll_ has joined #openstack-swift15:45
*** yarkot has joined #openstack-swift15:49
*** petertr7 is now known as petertr7_away15:51
*** Jeffrey4l has joined #openstack-swift15:53
*** jsuchome has left #openstack-swift15:53
*** dslevin has joined #openstack-swift15:54
*** doxavore has joined #openstack-swift15:54
*** aix has quit IRC15:55
doxavoreIf one would like to reduce the I/O load during replication (/rebalancing) to reduce impact on request latency, at the cost of longer replication times, would rsync's max connections be a good place to do that?15:55
doxavoreand if so, is there a recommended range you stay between?15:56
*** petertr7_away is now known as petertr715:58
*** haomaiwang has quit IRC16:01
*** mragupat has joined #openstack-swift16:01
*** haomaiwang has joined #openstack-swift16:01
*** dslevin has quit IRC16:06
*** garthb has joined #openstack-swift16:09
*** yatin has joined #openstack-swift16:13
*** shakamunyi has joined #openstack-swift16:14
*** barra204 has quit IRC16:15
*** diogogmt has joined #openstack-swift16:17
*** diazjf has joined #openstack-swift16:21
*** PsionTheory has joined #openstack-swift16:25
eranromacoles: thanks I am waiting :-)16:30
acoleseranrom: its looking good. i just have some suggestions to cleanup some of tests.16:31
eranromacoles: thanks. awaiting patiently :-)16:32
*** zhill has joined #openstack-swift16:33
*** robefran has quit IRC16:36
*** robefran has joined #openstack-swift16:38
*** aix has joined #openstack-swift16:43
*** diazjf has quit IRC16:56
*** ccontini has quit IRC16:57
acoleseranrom: over to you :)16:58
*** diazjf has joined #openstack-swift16:59
*** haomaiwang has quit IRC17:01
*** haomaiwang has joined #openstack-swift17:01
*** jmccarthy1 has quit IRC17:02
openstackgerritStuart McLaren proposed openstack/python-swiftclient: Retry download of object body  https://review.openstack.org/15920817:02
*** SkyRocknRoll has quit IRC17:03
*** gyee has joined #openstack-swift17:08
*** jistr has quit IRC17:19
mahaticacoles: I think master needs to be merged into crypto branch. patch 211451 failed jenkins, looks like it is not recognizing nosetests17:22
patchbotmahatic: https://review.openstack.org/#/c/211451/ - Modify unit tests to include real crypto17:22
*** petertr7 is now known as petertr7_away17:23
acolesmahatic: the unit test failed with AttributeError: 'module' object has no attribute 'get_crypto'17:25
mahaticacoles: ouch17:26
*** silor has quit IRC17:27
mahaticoh I don't have those tests with me yet, the round trip tests17:28
*** petertr7_away is now known as petertr717:29
*** yarkot has quit IRC17:30
*** yatin has quit IRC17:31
*** trifon has quit IRC17:32
*** rledisez has quit IRC17:35
*** joeljwright has quit IRC17:36
*** aix has quit IRC17:38
*** arnox has quit IRC17:45
*** mragupat has quit IRC17:56
*** mragupat has joined #openstack-swift17:56
*** jordanP has quit IRC17:59
openstackgerritMahati Chamarthy proposed openstack/swift: Modify unit tests to include real crypto  https://review.openstack.org/21145118:00
*** haomaiwang has quit IRC18:01
*** haomaiwang has joined #openstack-swift18:01
*** klrmn has joined #openstack-swift18:08
*** zaitcev has joined #openstack-swift18:09
*** ChanServ sets mode: +v zaitcev18:09
*** changbl has joined #openstack-swift18:12
*** hseipp has quit IRC18:15
*** yarkot has joined #openstack-swift18:21
*** Jeffrey4l has quit IRC18:24
openstackgerritAlistair Coles proposed openstack/python-swiftclient: Add functional test for object PUT with raw iterator  https://review.openstack.org/26383818:34
notmynamegood morning18:35
gmmahagood morning notmyname18:35
notmynamenot much happened in here during my overnight18:39
acolesnotmyname: we're all hard at work while you sleep :)18:40
* acoles will try to make more noise tomorrow18:40
*** acoles is now known as acoles_18:44
*** pdion891 has quit IRC18:54
*** haomaiwang has quit IRC19:01
*** haomaiwang has joined #openstack-swift19:01
*** yarkot has quit IRC19:05
*** yarkot has joined #openstack-swift19:12
openstackgerritOndřej Nový proposed openstack/swift: Fixed few misspellings in comments  https://review.openstack.org/26387019:21
*** dslevin has joined #openstack-swift19:25
openstackgerritOndřej Nový proposed openstack/python-swiftclient: Fixed few misspellings in comments  https://review.openstack.org/26387219:29
*** sarafraj has joined #openstack-swift19:34
*** minwoob has joined #openstack-swift19:38
onovynotmyname: thanks :]19:38
*** daemontool has quit IRC19:39
*** sarafraj has quit IRC19:41
*** daemontool has joined #openstack-swift19:41
*** ChubYann has joined #openstack-swift19:45
*** daemontool has quit IRC19:47
*** daemontool has joined #openstack-swift19:48
*** daemontool has quit IRC19:51
*** sarafraj has joined #openstack-swift19:57
*** sarafraj has left #openstack-swift19:58
*** haomaiwang has quit IRC20:01
*** haomaiwang has joined #openstack-swift20:01
ZyricGood Morning20:01
onovymorning? it's 9pm :)20:09
*** zhill has quit IRC20:09
*** bjkeller has joined #openstack-swift20:18
*** tsg has quit IRC20:24
*** gyee has quit IRC20:25
*** djc_ has joined #openstack-swift20:26
*** djc_ has quit IRC20:30
*** djc_ has joined #openstack-swift20:30
djc_I am getting a message when doing a swift stat: Account HEAD failed: https://x.x.x.x:8443/v1/AUTH_yyyy 503 Internal Server Error20:30
djc_not sure what to make of it.20:31
jrichlidjc_: I would check the logs if you could to see what the internal error was20:35
*** joeljwright has joined #openstack-swift20:37
*** ChanServ sets mode: +v joeljwright20:37
djc_jrichli: I checked the proxy logs and same type of message: http://pastebin.com/2se1bXED20:40
*** yarkot has quit IRC20:41
djc_jrichli: my proxy server is actually a lxc container. Not sure if that makes a difference. It shouldn't.20:43
jrichlidjc_: check the storage logs20:44
*** dslevin has quit IRC20:45
notmynamedjc_: take the trasaction id (txdf0d5292847e403f99ffa-00568c1fbf) and grep all the logs you have for it20:48
notmynamewell, you know it will be on your .215 and .214 server20:49
*** dslevin has joined #openstack-swift20:50
djc_notmyname: .214 and .215 are my storage nodes. I did grep the transaction id from /var/log/messages and it returned nothing. I am not seeing anything unusual on the storage nodes log other than a possible bad disk.20:52
openstackgerritBill Huber proposed openstack/swift: Re-format the SLO manifest file on new multipart-manifest GET call  https://review.openstack.org/26390220:56
*** mragupat has quit IRC20:57
notmynamedjc_: ah ok, then. for some reason your proxy server wasn't able to get a response from the account servers20:58
djc_notmyname: I am a relative novice to swift and I am reading the swift book you recommended a couple months back. The swift.conf and *.gz files is the same on all nodes20:58
notmynamedjc_: are you running with 3 replicas?20:58
*** mragupat has joined #openstack-swift20:58
djc_notmyname: 4 replicas20:58
djc_4 zones20:59
*** mragupat has quit IRC20:59
notmynameinterestingly, you don't have the same transaction id on those log messages21:00
*** haomaiwang has quit IRC21:01
*** haomaiwang has joined #openstack-swift21:01
*** mragupat has joined #openstack-swift21:01
djc_notmyname: looks like my ports are right for all the objects: http://pastebin.com/jY1iZ3Db21:02
notmynameyou can automate that check with `swift-recon --validate-servers`21:02
Guest79971clayg: why u no like my blueprint?!21:03
*** Guest79971 is now known as scotticus21:03
scotticusthats better.21:03
notmynameoh hi scotticus ;-)21:03
djc_notmyname: same with account server: http://pastebin.com/RGMae6Ay21:04
notmynamescotticus: are you referring to dev_id limits?21:04
scotticusthe 2^16 limit on devices, yes.21:05
*** mragupat has quit IRC21:05
*** mragupat has joined #openstack-swift21:05
*** mragupat has quit IRC21:06
*** mragupat has joined #openstack-swift21:06
notmynamescotticus: did you see any of the scrollback from yesterday when we talked about it in here?21:07
notmyname(if not I can summarize)21:07
notmynamethere were a few points21:07
scotticusthe summary i was relayed was clayg didn't think it was as valuable as fixing the rebalance.21:08
scotticusthat it "wasn't valuable"21:08
scotticusat least thats what was relayed to me.21:08
notmynamefirst is that the 65k limit is being hit (from what we're told) because of "holes" in the ring from older devices. so that should be fixed first instead of a 2**16 limit21:09
notmynamethat would allow more dev ids with less changes, and something that will benefit more deployers overall21:09
scotticusso that would bandaid the device limit temporarily...21:09
scotticusby freeing up some null devices.21:09
scotticusso since we're probably 1/2 way to that limit, i thought why not start working on somethign like this now.21:10
notmynamesecond IIRC is that there are some other things that would need to be worked on too (like rebalancing) to support these larger rings, and maybe that should be focused on first (trying to correctly summarize what clayg was saying)21:10
*** mragupat has quit IRC21:11
scotticuswhats the timeline for the rebalance changes?21:12
notmynameIMO (me, not clayg) I think it would be great to remove the 16 bit limit (and any limit). but it's a question of effort vs reward. eg filling the holes might be easier first. however pdardeau already is well along on removing the limit, so that's good21:12
scotticusis someone actively working on that?21:12
notmynameno, not AFAIK.21:12
scotticusis there a blueprint for the rebalance stuff?21:13
notmynameI think the discussion was not about if removing the limit was valuable, but what was valuable to do first. that was my takeaway21:13
notmynameI haven't seen pdardeau's patch yet, so I don't know what it involves. and I certainly want to ensure that larger policies keep working21:14
scotticushis patch is for removing the device size limit?21:14
notmynamebut basically having 30k+ drives in a ring comes with other problems, and the question was about those other things21:14
notmynameso IMO we should have a patch to fill the holes in the ring. and we should plan for removing the device limit altogether.21:15
notmynameIIRC clayg was only asking about the "why now" part. not if it was actually a good thing. just if it is a good thing now21:15
notmynamewhich I think was what got back to you21:15
scotticusprobably better now than later.21:16
*** zhill has joined #openstack-swift21:17
pdardeaunotmyname, scotticus: o/21:18
notmynameso that gets back around to the question that wasn't answered yesterday: how pressing is this? is it a Right Now thing? sometime this year? if current growth continues...21:18
notmynamepdardeau: did all that I just spewed make sense? line up with what you remember from yesterday?21:18
pdardeaunotmyname: yes, nice synopsis21:19
pdardeaunotmyname, scotticus: i can finish up device limit patch (a few more hours at most) and make it available for review21:19
pdardeauthen you guys can get a better feel for whether it's worth accepting at this time21:20
*** mragupat has joined #openstack-swift21:20
notmynameheh, if code's already written, it's much easier to accept ;-)21:20
pdardeauha, good point!21:21
*** chlong has quit IRC21:22
openstackgerritEran Rom proposed openstack/swift: Container-Sync to iterate only over synced containers  https://review.openstack.org/20580321:22
notmynamehowever, it's not like "put 100k devices in a ring" is likely to be a thing that only this patch is preventing. what about rebalancing? what about part power adjustments? what about ring-of-rings for better placements?21:22
pdardeaunotmyname: that's almost like a disturbance in the force, or something ;-)21:24
notmynamescotticus: does all that make sense? I don't remember anything from yesterday about removing or raising the limit being not valuable. just curious about why now and any plans around the other stuff that comes with rings that big21:24
scotticusall of it is good.  apparently pdardeau has done what the blueprint calls for.  the other stuff would absolutely need to get done at some point.21:25
notmynameand, hey. if pdardeau's got a patch, that's awesome. working code votes ;-)21:27
*** Vinsh has quit IRC21:29
*** petertr7 is now known as petertr7_away21:29
*** Vinsh has joined #openstack-swift21:30
*** mragupat has quit IRC21:30
*** gyee has joined #openstack-swift21:30
*** mragupat has joined #openstack-swift21:30
onovydoxavore: still here?21:31
*** mragupat_ has joined #openstack-swift21:32
onovydoxavore: https://github.com/openstack/swift/blob/master/etc/object-server.conf-sample#L172 this options do "shaping" on source side21:32
onovydoxavore: https://github.com/openstack/swift/blob/master/etc/rsyncd.conf-sample#L19 this on destination side21:33
*** Vinsh_ has joined #openstack-swift21:33
*** gyee has quit IRC21:34
*** mragupat has quit IRC21:35
*** Vinsh has quit IRC21:37
djc_notmyname: got any more advice for me?21:39
*** gyee has joined #openstack-swift21:41
*** yarkot has joined #openstack-swift21:41
notmynamedjc_: at this point you need to look on the account server and figure out why it's not responding to the proxy. you could test with a known object that goes to that server. or directly to the account server with curl21:47
notmynamedjc_: definitely need to make sure that the disk 14 is ok on that server21:47
notmynamedjc_: so I dont' have anything specific. just start going through the system to isolate the issue so you can figure out what it is21:48
djc_notmyname: ok thanks.21:48
*** petertr7_away is now known as petertr721:50
*** robefran has quit IRC21:52
*** dslevin has quit IRC21:54
*** cdelatte has quit IRC21:56
*** dslevin has joined #openstack-swift21:59
*** haomaiwang has quit IRC22:01
*** haomaiwang has joined #openstack-swift22:01
*** dslevin has quit IRC22:04
*** dslevin has joined #openstack-swift22:09
*** blmartin has quit IRC22:11
onovycan someone look to https://review.openstack.org/#/c/253038/ pls?22:22
*** haomaiwang has quit IRC22:23
*** haomaiwang has joined #openstack-swift22:28
hrouclayg, thanks a bunch for the comments re: symlink - I just wanted to validate:  So are we roughly on the same page that is post-as-copy is just about equally terrible with posting a regular object and a symlink (where we are redirecting the post to the target object i.e. the desired behavior) ... in the sense that we're not really introducing any additional consistency issues ?  If so we're going to push forward on evaluating fast-post in sy22:37
hrou(as I think there are some issues there we may need to resolve).22:37
*** petertr7 is now known as petertr7_away22:37
*** djc_ has quit IRC22:39
*** mragupat has joined #openstack-swift22:41
*** robefran has joined #openstack-swift22:41
mattoliverauhey hrou, happy new year! Haven't had a chance so say it to you yet :)22:43
hroumattoliverau, thanks a lot, and you too !22:43
*** mragupat_ has quit IRC22:44
openstackgerritAlan Erwin proposed openstack/swift: Refactoring the expiring objects feature  https://review.openstack.org/25208522:44
notmynamemattoliverau: just got my flights for LCA22:50
notmynameI'm not looking forward to LAX->MEL (15h45m)22:51
notmynamedo you take that one when you go home?22:51
doxavoreonovy: i'll work with those settings tonight and see where it lands. thanks!22:52
mattoliveraunotmyname: \o/22:58
mattoliveraunotmyname: anything thrrough LAX is annoying I prefer via SFO. Having said that, the last few times through LAX has actually been pretty good. I think it's the smart gates that has speed it up.. at least everytime I've been there lately22:59
mattoliveraubut yes, from SAT and YVR I came back on LAX > MEL23:00
gmmahahowdy mattoliverau23:00
*** daemontool has joined #openstack-swift23:00
mattoliveraugmmaha: hey man23:00
mattoliveraugmmaha: happy new year23:01
*** haomaiwang has quit IRC23:01
gmmahamattoliverau: happy new year to you too.. :)23:01
*** minwoob has quit IRC23:01
*** haomaiwang has joined #openstack-swift23:01
mattoliveraunotmyname: you coming in on Saturday before LCA or earlier?23:02
notmynameactually, just sending you an email about that :-)23:02
*** bjkeller has left #openstack-swift23:02
*** doxavore has quit IRC23:03
mattoliveraunotmyname: great, I will reply to the said email once I have recieved and read it :P23:04
*** joeljwright has quit IRC23:08
*** mragupat has quit IRC23:09
*** chlong has joined #openstack-swift23:12
*** diazjf has quit IRC23:12
*** chlong has quit IRC23:14
*** robefran has quit IRC23:15
*** kei_yama has joined #openstack-swift23:23
claygaerwin3_: did you *read* any of my comments on patch 252085 - you didn't respond to any of them?23:27
patchbotclayg: https://review.openstack.org/#/c/252085/ - Refactoring the expiring objects feature23:27
*** chlong has joined #openstack-swift23:30
mattoliverauhe may hav a clayg greese monkey filter :P23:34
*** daemontool has quit IRC23:40
claygmattoliverau: oh you published that!?23:40
*** robefran has joined #openstack-swift23:45
*** badari has quit IRC23:47
*** diogogmt has quit IRC23:54
*** ho has joined #openstack-swift23:58
*** ho is now known as Guest1444723:59

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