| *** gyee has quit IRC | 00:01 | |
| zaitcev | Let's see | 00:04 |
|---|---|---|
| zaitcev | go go gadget yum install pip | 00:06 |
| zaitcev | Package Arch Version Repository Size | 00:06 |
| zaitcev | ================================================================================ | 00:06 |
| zaitcev | Installing: | 00:06 |
| zaitcev | python-pip noarch 8.1.2-1.el7 openstack-queens 1.7 M | 00:06 |
| zaitcev | [root@keystone yum.repos.d]# rpm -qf /etc/yum.repos.d/rdo-release.repo | 00:08 |
| zaitcev | rdo-release-queens-2.noarch | 00:08 |
| zaitcev | That's about the extent of my knowledge on this topic. | 00:08 |
| *** renich has joined #openstack-swift | 00:18 | |
| *** renich has quit IRC | 02:10 | |
| *** early has quit IRC | 02:11 | |
| *** renich has joined #openstack-swift | 02:11 | |
| *** early has joined #openstack-swift | 02:30 | |
| *** psachin has joined #openstack-swift | 02:48 | |
| *** rcernin has quit IRC | 02:49 | |
| *** renich has quit IRC | 02:51 | |
| *** renich has joined #openstack-swift | 02:51 | |
| *** rcernin has joined #openstack-swift | 02:55 | |
| timburke | funny -- i see it in http://mirror.centos.org/centos/7/cloud/x86_64/openstack-queens/Packages/p/ but not the corresponding ones for rocky through train | 02:57 |
| *** renich has quit IRC | 03:05 | |
| *** rcernin has quit IRC | 03:08 | |
| *** psachin has quit IRC | 03:31 | |
| *** psachin has joined #openstack-swift | 03:33 | |
| *** rcernin has joined #openstack-swift | 03:45 | |
| *** rcernin has quit IRC | 03:45 | |
| *** rcernin has joined #openstack-swift | 03:50 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Install pip from upstream https://review.opendev.org/735992 | 04:03 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Use ensure-pip role https://review.opendev.org/735992 | 04:31 |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-swift | 04:33 | |
| timburke | oh. https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/templates/centos7/etc/yum.repos.d/epel.repo.j2#L6 | 04:34 |
| timburke | ok, iiuc once https://review.opendev.org/#/c/735523/ lands, we can recheck https://review.opendev.org/#/c/735992/ and *then* the gate will actually be in a state where we can land code again | 05:27 |
| patchbot | patch 735523 - devstack (stable/ussuri) - Fix uwsgi issues - 6 patch sets | 05:27 |
| patchbot | patch 735992 - swift - Use ensure-pip role - 5 patch sets | 05:27 |
| *** threestrands has joined #openstack-swift | 05:27 | |
| *** threestrands has quit IRC | 05:33 | |
| *** m75abrams has joined #openstack-swift | 06:00 | |
| *** rpittau|afk is now known as rpittau | 06:56 | |
| *** ccamacho has joined #openstack-swift | 07:12 | |
| *** psachin has quit IRC | 07:15 | |
| *** psachin has joined #openstack-swift | 07:17 | |
| *** zaitcev has quit IRC | 07:42 | |
| *** zaitcev has joined #openstack-swift | 07:46 | |
| *** ChanServ sets mode: +v zaitcev | 07:46 | |
| *** rcernin has quit IRC | 07:47 | |
| *** tkajinam has quit IRC | 08:57 | |
| *** cschwede has joined #openstack-swift | 09:21 | |
| *** ChanServ sets mode: +v cschwede | 09:21 | |
| *** rpittau is now known as rpittau|bbl | 10:18 | |
| *** psachin has quit IRC | 10:31 | |
| *** psachin has joined #openstack-swift | 11:03 | |
| *** psachin has quit IRC | 11:58 | |
| *** psachin has joined #openstack-swift | 12:00 | |
| *** rpittau|bbl is now known as rpittau | 12:06 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Add py3 probe tests on CentOS 8 https://review.opendev.org/690717 | 13:24 |
| *** rpittau is now known as rpittau|brb | 13:39 | |
| *** rpittau|brb is now known as rpittau | 13:53 | |
| *** psachin has quit IRC | 14:34 | |
| *** manuvakery has joined #openstack-swift | 14:46 | |
| *** m75abrams has quit IRC | 14:56 | |
| *** josephillips has joined #openstack-swift | 15:01 | |
| josephillips | Hey guys | 15:01 |
| josephillips | im reading documentation today and yesterday and im trying to understand something. | 15:02 |
| josephillips | what is the relation between part-power is per object per account per container? | 15:03 |
| josephillips | or exist this relation | 15:03 |
| josephillips | or i can use a number very high for part-power without side effect ? | 15:03 |
| josephillips | just a couples mores of mb | 15:03 |
| josephillips | in memory? | 15:04 |
| *** gyee has joined #openstack-swift | 15:58 | |
| DHE | josephillips: it's a setting that is per-ring file. so accounts, containers, and each object storage policy can have its own part power | 16:11 |
| DHE | The standard recommendation is that each disk have ~100 partitions assigned to it, biasing for more than 100 rather than less than 100 | 16:12 |
| DHE | higher values do use more RAM, but it also slows down replication | 16:13 |
| timburke | note that "use more RAM" covers more than just the in-memory ring data structure -- though that's a factor, the bigger problem (in my mind, anyway) is the increased number of inodes the backend server needs to deal with, and we want to have those in RAM whenever possible | 16:16 |
| *** rpittau is now known as rpittau|afk | 16:17 | |
| timburke | if you really want to dive into the nitty-gritty of partitions and part power, https://docs.openstack.org/swift/latest/ring_background.html is a great starting point | 16:19 |
| DHE | one thing I think I noticed is that, say I launch 16 proxy server processes. they all share the initial ring file which was loaded before the fork()ing. but ring changes causes all 16 to relaod the ring and memory usage explodes | 16:20 |
| DHE | shared memory shenanigans | 16:20 |
| tdasilva | josephillips: here's another very good deep dive resource: https://www.youtube.com/watch?v=ger20cqOypE&t | 16:22 |
| *** cschwede has quit IRC | 16:33 | |
| josephillips | tdasilva: yeah i saw this last night | 16:59 |
| josephillips | i understand how works but im still dont found the relation between part-power | 17:00 |
| josephillips | because you use /srv/node/disk/PART/ | 17:00 |
| josephillips | what is the factor that swift choose use this partition | 17:00 |
| josephillips | can get full this partition? | 17:01 |
| josephillips | what is the factor that swift said ok . i dont going to use this paritiion and i will use a new one | 17:01 |
| josephillips | the quantity of objects | 17:02 |
| DHE | it's /srv/node/mountpoint/ring-id/partno | 17:04 |
| DHE | all partitions on a device share the same mount point, so the device will fill up and all partitions allocated to it would be full | 17:05 |
| DHE | don't confuse a disk partition with a swift ring partition | 17:05 |
| josephillips | DHE: i know but what is the factor of use a new partition | 17:14 |
| DHE | I don't understand the question | 17:15 |
| josephillips | /srv/node/mountpoint/ring-id/partno | 17:20 |
| josephillips | here | 17:20 |
| josephillips | what swift do to choose these partno | 17:20 |
| josephillips | > | 17:20 |
| josephillips | and when swift decide to use it or not | 17:20 |
| *** josephillips has quit IRC | 17:35 | |
| timburke | so the partition is just a directory on disk -- it's a group of objects whose name have some hash bits in common | 17:43 |
| timburke | a filename for object data might look something like /srv/node3/sdb7/objects/109/05c/1b6b491073cb8fbd749addce3cf9205c/1592415748.91691.data | 17:43 |
| timburke | 1b6b491073cb8fbd749addce3cf9205c -> hash; this is something like md5(hash_prefix + acct/cont/obj + hash_suffix) | 17:44 |
| timburke | 1592415748.91691 -> timestamp; in case of overwrite, oldest timestamp wins (eventually) | 17:45 |
| timburke | 05c -> suffix; this is just the last 3 chars from the hash | 17:46 |
| timburke | 109 -> partition; this is the first part-number bits from the hash, taken as a decimal number | 17:47 |
| timburke | my part-power is 10, so that gets me 1b and partway through 6: 0b0001101101 | 17:49 |
| timburke | all objects whose hash start 1b4 through 1b7 will be in the same partition | 17:52 |
| openstackgerrit | Merged openstack/swift master: Use ensure-pip role https://review.opendev.org/735992 | 18:00 |
| timburke | gate's working again! \o/ | 18:08 |
| DHE | Oldest timestamp? | 18:11 |
| *** josephillips has joined #openstack-swift | 18:14 | |
| timburke | bah. newest, of course. i don't know why i said that | 18:18 |
| tdasilva | timburke: I think josephillips missed your explanation | 18:18 |
| tdasilva | FWIW http://eavesdrop.openstack.org/irclogs/%23openstack-swift/latest.log.html | 18:19 |
| timburke | http://eavesdrop.openstack.org/irclogs/%23openstack-swift/%23openstack-swift.2020-06-17.log.html#t2020-06-17T17:43:08 even, if you want to skip right to it ;-) | 18:20 |
| *** josephillips has quit IRC | 18:22 | |
| *** manuvakery has quit IRC | 18:34 | |
| *** ktsuyuzaki has joined #openstack-swift | 20:11 | |
| *** ChanServ sets mode: +v ktsuyuzaki | 20:11 | |
| *** cwright_ has joined #openstack-swift | 20:11 | |
| *** ktsuyuzaki is now known as kota__ | 20:12 | |
| *** irclogbot_0 has quit IRC | 20:14 | |
| *** gyee has quit IRC | 20:17 | |
| timburke | clayg, any chance you'd have some cycles to look at https://review.opendev.org/#/c/727845/ again? i'd like to get a release together, but i think i'd prefer if we had that... | 20:18 |
| patchbot | patch 727845 - python-swiftclient - Application credential support follow-up - 4 patch sets | 20:18 |
| *** cwright has quit IRC | 20:18 | |
| *** kota_ has quit IRC | 20:18 | |
| timburke | i also need to think about what to do about https://review.opendev.org/#/c/721051/ :-( | 20:19 |
| patchbot | patch 721051 - python-swiftclient - Only add __del__ to HTTPConnection shim on py3 - 2 patch sets | 20:19 |
| *** irclogbot_1 has joined #openstack-swift | 20:21 | |
| clayg | timburke: I'm fine with p 727845 - do you know of ANYONE else that uses openstackclient? Or have a link/tip on how to install it (is it python?) | 20:31 |
| patchbot | https://review.opendev.org/#/c/727845/ - python-swiftclient - Application credential support follow-up - 4 patch sets | 20:31 |
| *** gyee has joined #openstack-swift | 20:31 | |
| clayg | i don't suppose we can ignore py2.7 and hope p 721051 goes away? | 20:33 |
| patchbot | https://review.opendev.org/#/c/721051/ - python-swiftclient - Only add __del__ to HTTPConnection shim on py3 - 2 patch sets | 20:33 |
| *** nottrobin has quit IRC | 20:36 | |
| timburke | clayg, it's python and it's listed in our test-requirements.txt, so i'm willing to bet you've already got it in your saio ;-) | 20:36 |
| timburke | as for swiftclient and py2 -- at the very least, i'd prefer intentionally having a final py2 release over quietly until things get so horribly broken that py2 is unsustainable -- but y'know, we'll do what we have to i suppose | 20:38 |
| *** abelur has quit IRC | 20:40 | |
| timburke | (as much as anything, that's so we're sure to appropriately tag supported python versions so there isn't a final FU to people stuck on py2 trying to install via pip) | 20:40 |
| clayg | DO NOT type openstack --help w/o a large scrollback buffer 🤣 | 20:44 |
| timburke | almost meeting time! | 20:57 |
| kota__ | it seems like that my bouncer got problem so I changed my nick temporarly | 21:01 |
| *** noonedeadpunk has quit IRC | 21:12 | |
| *** nottrobin has joined #openstack-swift | 21:30 | |
| *** abelur has joined #openstack-swift | 21:40 | |
| clayg | timburke: i'm about to sign off - i'll leave my re-complete going over night for p 733919 - but i may not get to looking up the abort stuff until later | 22:03 |
| patchbot | https://review.opendev.org/#/c/733919/ - swift - s3api: Allow CompleteMultipartUpload requests to b... - 3 patch sets | 22:03 |
| clayg | we should have asked about the config option - maybe it'll be useful to know if it's like maybe predictably 24 hours or something on aws | 22:03 |
| clayg | but even then, maybe it makes more sense to just change the default than drop the config option? it could be more like a constraint maybe? | 22:04 |
| timburke | oh, there's an idea for another approach to things like max_bucket_listing, max_parts_listing, max_upload_part_num, max_multi_delete_objects, min_segment_size... add a [s3-constraints] section to swift.conf... | 22:06 |
| openstackgerrit | Merged openstack/python-swiftclient master: Application credential support follow-up https://review.opendev.org/727845 | 22:41 |
| *** rcernin has joined #openstack-swift | 22:42 | |
| *** seongsoocho has quit IRC | 22:50 | |
| *** seongsoocho has joined #openstack-swift | 22:51 | |
| *** gyee has quit IRC | 22:52 | |
| *** tkajinam has joined #openstack-swift | 22:54 | |
| *** hoonetorg has quit IRC | 23:10 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Close connections created when calling module-level functions https://review.opendev.org/721051 | 23:14 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Revert "Cleanup session on delete" https://review.opendev.org/736401 | 23:14 |
| *** hoonetorg has joined #openstack-swift | 23:22 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!