| *** renich has joined #openstack-swift | 00:42 | |
| *** gyee has quit IRC | 00:57 | |
| openstackgerrit | Merged openstack/swift master: Clean up account-reaper a bit https://review.openstack.org/611686 | 01:05 |
|---|---|---|
| openstackgerrit | Merged openstack/swift master: s3token: Add note about config change when upgrading from swift3 https://review.openstack.org/641153 | 01:07 |
| *** mikecmpbll has quit IRC | 01:21 | |
| *** renich has quit IRC | 01:48 | |
| *** renich has joined #openstack-swift | 01:51 | |
| *** psachin has joined #openstack-swift | 02:57 | |
| *** itlinux has joined #openstack-swift | 03:04 | |
| openstackgerrit | Merged openstack/swift master: Simplify empty suffix handling https://review.openstack.org/637662 | 03:06 |
| openstackgerrit | Merged openstack/swift master: Refactor write_affinity DELETE handling https://review.openstack.org/641527 | 03:11 |
| openstackgerrit | Merged openstack/swift master: Implemented the fix to handle the HTTP request methods other than GET. https://review.openstack.org/566806 | 03:13 |
| *** tkajinam has quit IRC | 03:54 | |
| *** tkajinam has joined #openstack-swift | 03:57 | |
| notmyname | kota_: around? | 03:58 |
| *** psachin has quit IRC | 04:34 | |
| *** psachin has joined #openstack-swift | 04:42 | |
| *** itlinux has quit IRC | 04:45 | |
| *** sleterrier_ has joined #openstack-swift | 05:35 | |
| *** sleterrier_ has quit IRC | 05:38 | |
| *** _david_sohonet has quit IRC | 05:59 | |
| *** renich has quit IRC | 06:07 | |
| *** renich_ has joined #openstack-swift | 06:07 | |
| *** renich_ is now known as renich | 06:09 | |
| *** renich has quit IRC | 06:54 | |
| *** e0ne has joined #openstack-swift | 07:17 | |
| *** pcaruana has joined #openstack-swift | 07:21 | |
| *** e0ne has quit IRC | 07:27 | |
| *** rchurch has joined #openstack-swift | 07:40 | |
| *** e0ne has joined #openstack-swift | 07:42 | |
| *** ccamacho has joined #openstack-swift | 08:11 | |
| *** e0ne has quit IRC | 08:15 | |
| *** tkajinam has quit IRC | 08:35 | |
| *** rcernin has quit IRC | 08:38 | |
| *** mikecmpbll has joined #openstack-swift | 09:23 | |
| *** psachin has quit IRC | 09:48 | |
| *** psachin has joined #openstack-swift | 10:25 | |
| *** e0ne has joined #openstack-swift | 10:45 | |
| *** dlcastor has joined #openstack-swift | 11:12 | |
| dlcastor | hello everyone; I'm trying to "empty" a disk so I can then replace it. For that, I did a swift-ring-builder on all 3 ring with set_weight id 0 on a cluster node and then I ran rebalance on all the rings. Then, I copied the ring.gz files to the other nodes. Now, on the other nodes I get Ring file /etc/swift/object.ring.gz is obsolete for all rings - any ideas? | 11:14 |
| *** ccamacho has quit IRC | 11:24 | |
| *** ccamacho has joined #openstack-swift | 11:26 | |
| *** rcernin has joined #openstack-swift | 11:28 | |
| *** pcaruana has quit IRC | 11:38 | |
| *** e0ne has quit IRC | 12:07 | |
| *** psachin has quit IRC | 12:11 | |
| *** pcaruana has joined #openstack-swift | 12:16 | |
| *** psachin has joined #openstack-swift | 12:19 | |
| tdasilva | dlcastor: having a difficult time understanding your description, when do you get that error? after copying rings to other nodes. From looking at swift code, the only place that you would get a similar message would be here: https://github.com/openstack/swift/blob/master/swift/cli/ringbuilder.py#L517 | 12:23 |
| tdasilva | so it seems that you get that message when you are actually running the ringbuilder, but you may have a older .builder file lying around (or vice-versa) | 12:24 |
| dlcastor | sorry for not being clear | 12:49 |
| dlcastor | i have 3 nodes, on one node I ran: | 12:49 |
| dlcastor | swift-ring-builder /etc/swift/container.builder set_weight d18 0 (and then object and account) | 12:49 |
| dlcastor | swift-ring-builder /etc/swift/container.builder rebalance (and then object and account) | 12:49 |
| dlcastor | then I copied the /etc/swift/account.ring.gz container.ring.gz object.ring.gz to the other 2 nodes | 12:49 |
| dlcastor | now, when I run swift-ring-builder /etc/swift/account.builder object.builder or account.builder on the 2 nodes where I copied the rings.gz I get | 12:50 |
| dlcastor | swift-ring-builder /etc/swift/account.builder | 12:50 |
| dlcastor | /etc/swift/account.builder, build version 33 | 12:50 |
| dlcastor | 1024 partitions, 3.000000 replicas, 1 regions, 1 zones, 30 devices, 0.59 balance, 0.00 dispersion | 12:50 |
| dlcastor | The minimum number of hours before a partition can be reassigned is 1 (0:00:00 remaining) | 12:50 |
| dlcastor | The overload factor is 0.00% (0.000000) | 12:50 |
| dlcastor | Ring file /etc/swift/account.ring.gz is obsolete | 12:50 |
| dlcastor | so, you are saying that the account, object and container builder files on my other 2 nodes are older than the ones on the node I ran the set-weight and rebalance? then, how can I update those? | 12:51 |
| tdasilva | dlcastor: so yeah, i think i understand the confusion. you should only run swift-ring-builder in one machine, that's typically where you would do all your ring changes. It doesn't have to be a node in the cluster, could just be your host machine. Once you are doing the rebalance then you need to copy the ring.gz file to the cluster nodes | 13:12 |
| tdasilva | dlcastor: but make sure you backup your .builder files, you never want to lose them. One way people typically backup .builder files is by copying them together with the .ring.gz files to the cluster nodes. In that case, you just have to make sure that you are always copying all 6 files (in case only one storage policy) 3 .builder and 3 .ring.gz | 13:15 |
| *** rcernin has quit IRC | 13:16 | |
| *** e0ne has joined #openstack-swift | 13:16 | |
| *** ccamacho has quit IRC | 13:23 | |
| *** ccamacho has joined #openstack-swift | 13:23 | |
| *** ccamacho has quit IRC | 13:28 | |
| *** ccamacho has joined #openstack-swift | 13:28 | |
| *** psachin has quit IRC | 13:30 | |
| dlcastor | tdasilva: so the 6 files are required _only_ when you are operating on the cluster and it's best practice to have them backed up on all the nodes with the latest version; did I get that right? | 14:00 |
| tdasilva | dlcastor: you make changes to the rings using .builder files (i don't think *.ring.gz files are even required to be present). *.ring.gz are just static output, can't make changes to them. The cluster only needs the *.ring.gz to properly function (i.e., .builder are not required on cluster nodes). You can backup .builder files anyway you would like, one practice is to copy them while also copying .ring.gz files to | 14:04 |
| tdasilva | cluster nodes | 14:04 |
| tdasilva | dlcastor: i really just wanted to clarify that you run swift-ring-builder on just one node, not on every cluster node... | 14:06 |
| tdasilva | dlcastor: https://docs.openstack.org/swift/latest/admin_guide.html#managing-the-rings | 14:07 |
| dlcastor | I run the swift-ring-builder with set-weight and rebalance _only_ on one node | 14:18 |
| dlcastor | on the others, I ran it without doing anything, like just swift-ring-builder builder_file | 14:19 |
| *** itlinux has joined #openstack-swift | 14:39 | |
| tdasilva | dlcastor: ah, understood, you were trying to just see the output. so yeah, in that case it's just because you had an older .builder file on that node... | 14:40 |
| dlcastor | tdasilva: thanks a lot for your help and explanations ;-) | 14:40 |
| tdasilva | yw, good luck | 14:41 |
| openstackgerrit | Alexandre Lécuyer proposed openstack/swift feature/losf: WIP - start working on ansible playbooks for LOSF https://review.openstack.org/644879 | 14:45 |
| *** mvkr has quit IRC | 15:30 | |
| *** renich has joined #openstack-swift | 15:32 | |
| *** renich has quit IRC | 15:48 | |
| *** gyee has joined #openstack-swift | 15:51 | |
| *** itlinux has quit IRC | 16:00 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Fix how we UTF-8-ify func tests https://review.openstack.org/641855 | 16:00 |
| *** mvkr has joined #openstack-swift | 16:07 | |
| *** itlinux has joined #openstack-swift | 16:09 | |
| *** itlinux has quit IRC | 16:14 | |
| *** itlinux has joined #openstack-swift | 16:17 | |
| *** renich has joined #openstack-swift | 16:21 | |
| *** itlinux has quit IRC | 16:27 | |
| *** itlinux has joined #openstack-swift | 16:29 | |
| *** itlinux has quit IRC | 16:37 | |
| *** altlogbot_2 has joined #openstack-swift | 16:40 | |
| *** altlogbot_2 has quit IRC | 16:41 | |
| *** itlinux has joined #openstack-swift | 16:43 | |
| *** itlinux has quit IRC | 16:50 | |
| notmyname | I've got a customer meeting to be at this afternoon, so timburke will be leading today's meeting | 17:07 |
| *** e0ne has quit IRC | 17:07 | |
| *** altlogbot_1 has joined #openstack-swift | 17:27 | |
| *** renich has quit IRC | 17:28 | |
| *** altlogbot_1 has quit IRC | 17:28 | |
| *** renich_ has joined #openstack-swift | 17:29 | |
| *** altlogbot_0 has joined #openstack-swift | 17:30 | |
| clayg | @timburke if you wanna see p 633671 move forward can we at least confirm lp bug #1813725 if not set a priority? | 17:31 |
| openstack | Launchpad bug 1813725 in OpenStack Object Storage (swift) "Object encryption key truncated sometimes when used with Barbican" [Undecided,New] https://launchpad.net/bugs/1813725 | 17:31 |
| patchbot | https://review.openstack.org/#/c/633671/ - swift - Fix decryption for broken objects - 8 patch sets | 17:31 |
| *** mvkr has quit IRC | 17:31 | |
| *** ccamacho has quit IRC | 17:36 | |
| *** mikecmpbll has quit IRC | 17:59 | |
| *** renich_ has quit IRC | 18:14 | |
| *** renich_ has joined #openstack-swift | 18:17 | |
| *** e0ne has joined #openstack-swift | 18:17 | |
| *** irclogbot_2 has quit IRC | 18:21 | |
| *** irclogbot_3 has joined #openstack-swift | 18:24 | |
| *** irclogbot_3 has quit IRC | 18:27 | |
| *** irclogbot_0 has joined #openstack-swift | 18:29 | |
| *** clayg has quit IRC | 18:30 | |
| openstackgerrit | Merged openstack/swift master: Fix how we UTF-8-ify func tests https://review.openstack.org/641855 | 18:30 |
| *** clayg has joined #openstack-swift | 18:33 | |
| *** ChanServ sets mode: +v clayg | 18:33 | |
| *** mikecmpbll has joined #openstack-swift | 18:35 | |
| kota_ | hello | 18:43 |
| kota_ | timburke: I might not be available today's meeting because of business meeting. | 18:44 |
| *** e0ne has quit IRC | 18:53 | |
| *** zaitcev_ is now known as zaitcev | 18:56 | |
| zaitcev | That Michele Valsecchi is pretty good, as it turns out. Although her initial contribution was to fix typo. I'm wondering if we could entice her to do more. | 18:57 |
| *** altlogbot_0 has quit IRC | 19:00 | |
| *** altlogbot_1 has joined #openstack-swift | 19:02 | |
| *** dlcastor has quit IRC | 19:04 | |
| *** altlogbot_1 has quit IRC | 19:06 | |
| *** altlogbot_0 has joined #openstack-swift | 19:08 | |
| *** irclogbot_0 has quit IRC | 19:16 | |
| *** irclogbot_2 has joined #openstack-swift | 19:18 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: Rename ContainerBroker.path to cpath https://review.openstack.org/635677 | 19:19 |
| *** irclogbot_2 has quit IRC | 19:30 | |
| *** irclogbot_1 has joined #openstack-swift | 19:32 | |
| *** altlogbot_0 has quit IRC | 19:34 | |
| *** altlogbot_2 has joined #openstack-swift | 19:36 | |
| clayg | timburke: why did we think that hardlinks should make listings have a hardlink_size key (instead of just overwriting "bytes")? | 19:41 |
| clayg | timburke: I don't feel like there's really a "loss of information" by replacing the 0 with something more interesting... | 19:41 |
| clayg | we decided for content-type we can just let the symlink have the content-type of the target object resp | 19:42 |
| clayg | i know in versioned writes we were going to fix the etag - so if we do go with symlink_bytes we can fix it there too, but it seems maybe useful to just do it for all hardlinks? | 19:44 |
| timburke | clayg, double-counting bytes in the container stats | 20:45 |
| *** irclogbot_1 has quit IRC | 20:45 | |
| *** mikecmpbll has quit IRC | 20:46 | |
| clayg | no, i just mean in the *listings* like SLO does - in the row bytes i'd still have zero - i'm talking like what SLO does | 20:47 |
| *** mikecmpbll has joined #openstack-swift | 20:47 | |
| *** irclogbot_3 has joined #openstack-swift | 20:47 | |
| clayg | we add the new symlink_etag key (using the symlink_target_etag value from the etag column), for some reason I wrote down symlink_size key too - but I think really I can just overwrite bytes? | 20:47 |
| clayg | ... but that's what I'm double-checking ... either way there's no double counting (that I'm aware of!) | 20:48 |
| timburke | ...'cause i hate the way SLO does it? *shrug* | 20:48 |
| clayg | orly? the only thing I remember hating about SLO is that it uses content-type | 20:48 |
| clayg | well I guess a SLO manifest *does* have a valid (useful?) value in bytes already - so that one IS lossy | 20:49 |
| timburke | that's the big reason i hate it. the other part is that the cotnainer-server needs to know about it, and do the translation there (so m/w never even get a chance to second-guess it) | 20:49 |
| timburke | maybe since symlinks are by definition zero-bytes *in actuality*, it's fine? idk | 20:50 |
| *** e0ne has joined #openstack-swift | 20:52 | |
| *** altlogbot_2 has quit IRC | 20:53 | |
| clayg | well, doing it in the container server is just silly 🤣 | 20:54 |
| *** altlogbot_2 has joined #openstack-swift | 20:55 | |
| timburke | so why *not* do it the better way? establish precedent that was actually like, and hope that some day we get data migrations to fix up on-disk SLOs? | 20:57 |
| clayg | ok, well it turns out it's pretty easy to flip that around either way | 20:57 |
| timburke | meeting ping clayg tdasilva zaitcev cschwede alecuyer rledisez kota_ mattoliverau | 20:58 |
| clayg | timburke: sure, we can return symlink_bytes & bytes - i'm glad I ran it by you! | 20:58 |
| timburke | (notmyname's in a meeting, so i'll chair) | 20:58 |
| mattoliverau | morning, I was just in a OUI meeting re:denver. done now | 20:59 |
| notmyname | oh hi | 21:31 |
| timburke | oh, right! i should create an etherpad for topics for Denver for next week... | 21:33 |
| *** itlinux has joined #openstack-swift | 21:39 | |
| *** mikecmpbll has quit IRC | 21:59 | |
| zaitcev | What?! Denver is still a month away, ISN'T IT | 22:00 |
| zaitcev | https://www.linkedin.com/jobs/view/1073741562/ | 22:00 |
| *** e0ne has quit IRC | 22:00 | |
| zaitcev | (nowhere in the ad is mentioned if RAX is going to terminate the whole project in a few months as they did with Hummingbird) | 22:01 |
| timburke | hey, better to start collecting ideas early! | 22:01 |
| notmyname | zaitcev: yeah, I wouldn't recommend that anyone actively purse jobs at rackspace at this point. at least based on all their recent layoffs | 22:08 |
| *** pcaruana has quit IRC | 22:11 | |
| *** mvkr has joined #openstack-swift | 22:16 | |
| *** rcernin has joined #openstack-swift | 22:57 | |
| *** rcernin has quit IRC | 22:58 | |
| *** rcernin has joined #openstack-swift | 22:58 | |
| timburke | notmyname just reminded me how https://review.openstack.org/#/c/629659/ might be a good thing to get... | 22:59 |
| patchbot | patch 629659 - swift - slo_manifest_hook follow-up - 2 patch sets | 22:59 |
| *** itlinux has quit IRC | 22:59 | |
| notmyname | I'M NOT LOOKING FOR MORE PATCHES TO LAND, TIM | 22:59 |
| *** itlinux has joined #openstack-swift | 23:00 | |
| *** itlinux has quit IRC | 23:08 | |
| *** itlinux has joined #openstack-swift | 23:09 | |
| *** renich has joined #openstack-swift | 23:16 | |
| *** renich_ has quit IRC | 23:18 | |
| openstackgerrit | Merged openstack/swift master: Verify client input for v4 signatures https://review.openstack.org/629301 | 23:52 |
| openstackgerrit | Merged openstack/swift master: Clean up how we walk through ranges in ECAppIter https://review.openstack.org/638019 | 23:56 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!