| *** mikecmpbll has quit IRC | 00:54 | |
| *** two_tired has joined #openstack-swift | 01:18 | |
| kota_ | clayg: alright, I'll visit the patch today, thanks for bringing that. | 04:08 |
|---|---|---|
| *** two_tired has quit IRC | 04:45 | |
| openstackgerrit | Merged openstack/swift master: py3: object server https://review.openstack.org/629610 | 05:49 |
| *** [diablo] has quit IRC | 06:04 | |
| *** [diablo] has joined #openstack-swift | 06:19 | |
| *** itlinux_ has joined #openstack-swift | 06:28 | |
| *** itlinux has quit IRC | 06:30 | |
| *** itlinux_ has quit IRC | 06:48 | |
| *** rcernin has quit IRC | 06:59 | |
| *** e0ne has joined #openstack-swift | 07:30 | |
| *** hseipp has joined #openstack-swift | 07:42 | |
| kota_ | Oh nice. py3 object-server has been coming. | 07:43 |
| *** ccamacho has joined #openstack-swift | 07:56 | |
| *** e0ne has quit IRC | 08:04 | |
| *** ccamacho has quit IRC | 08:25 | |
| *** mikecmpb_ has joined #openstack-swift | 09:06 | |
| *** ccamacho has joined #openstack-swift | 09:08 | |
| *** pcaruana has joined #openstack-swift | 09:30 | |
| *** e0ne has joined #openstack-swift | 09:38 | |
| *** jungleboyj has quit IRC | 09:39 | |
| *** jungleboyj has joined #openstack-swift | 09:42 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: Save original index in the determine_chunk_destinations test https://review.openstack.org/630934 | 11:10 |
| kota_ | clayg: ^ it's small fix an issue I found while checking the frag_index specification. | 11:10 |
| kota_ | small fix for an issue | 11:10 |
| *** openstackgerrit has quit IRC | 11:22 | |
| *** hseipp has quit IRC | 12:33 | |
| *** gyee has joined #openstack-swift | 12:53 | |
| *** gyee has quit IRC | 13:34 | |
| *** mikecmpb_ is now known as mikecmpbll | 13:51 | |
| *** mvkr has quit IRC | 13:58 | |
| *** psachin has joined #openstack-swift | 14:16 | |
| *** mvkr has joined #openstack-swift | 15:00 | |
| timburke | good morning | 15:56 |
| *** e0ne has quit IRC | 16:11 | |
| *** ccamacho has quit IRC | 16:16 | |
| *** mvkr has quit IRC | 16:56 | |
| *** khelegond has joined #openstack-swift | 17:09 | |
| *** e0ne has joined #openstack-swift | 17:13 | |
| notmyname | good morning | 17:19 |
| khelegond | 'morning! | 17:22 |
| khelegond | Well, will use this opportunity to ask a quick question. I'm trying to shard a large container in my swift cluster. I know I need to use swift-manage-shard-ranges, but I can't seem to find the exact place where the container_db is. | 17:24 |
| notmyname | have you used `swift-get-nodes <container ring> <account> <container>`? | 17:24 |
| khelegond | I thought I was supposed to use swift-get-nodes to find where they are, but... the command is showing me directories that simply doesn't exist in the machine | 17:25 |
| notmyname | make sure you're using the right ring | 17:25 |
| khelegond | sudo swift-get-nodes -a /etc/swift/container.ring.gz <account> <container> | 17:25 |
| khelegond | If that's what I'm supposed to be doing, then I'll dig a little deeper :) | 17:26 |
| notmyname | looks correct | 17:27 |
| khelegond | Ok then - I'll try to understand why my get-nodes is showing me weird data. Thanks! | 17:27 |
| notmyname | check that the ring is correct. eg validate it with swift-recon on all your nodes and make sure they have the same md5 | 17:28 |
| *** e0ne has quit IRC | 17:28 | |
| khelegond | Perfect! Yeah, I'm kinda thinking that there's something weird with the rings. | 17:28 |
| khelegond | Thanks once again! | 17:28 |
| notmyname | could be that you've got a typo in the account or container name. if it's got stuff that needs to be escaped, you'll need to account for that too. | 17:29 |
| khelegond | Yeah - but I chose simple account and container names to test the functionality before moving to production. So the names are pretty standard. Will look around more! | 17:30 |
| timburke | if you've already got the sharders running, you should be able to look in container recon dumps to find large containers | 17:31 |
| khelegond | Oh - where do they dump this info? | 17:31 |
| timburke | that'll find large DBs even if they're stuck on handoffs | 17:31 |
| timburke | check recon_cache_path -- it'll default to /var/cache/swift, then there should be a container.recon under there | 17:32 |
| khelegond | Great, thanks!!! | 17:33 |
| notmyname | timburke: is that not exposed via an option to `swift-recon`? | 17:33 |
| timburke | not sure, which means probably not :-( | 17:33 |
| timburke | khelegond: you'll want to check on all nodes, of course | 17:34 |
| *** mikecmpbll has quit IRC | 17:34 | |
| khelegond | Yeah - doing this right now | 17:34 |
| timburke | it's JSON formatted; something like `< /var/cache/swift/container.recon jq .sharding_stats.sharding.sharding_candidates` should get you what you're looking for | 17:36 |
| timburke | but note that it'll only include containers it finds over the shard_container_threshold | 17:36 |
| khelegond | Yeah just found it. It'll define based on "shard_container_threshold", on the conf, right? However, I thought that this option would only matter if auto-sharding was enabled. | 17:37 |
| timburke | bah; i think we forgot to update the comments in the sample conf. we got nervous about completely automated sharding, so we made this manual trigger... then we realized we need some way to know that we ought to pull the trigger... | 17:39 |
| khelegond | Great! | 17:42 |
| khelegond | So I'll define the value on the conf, and wait for the recon to show me which containers I can shard. Thanks! | 17:42 |
| *** pcaruana has quit IRC | 18:01 | |
| *** psachin has quit IRC | 18:03 | |
| khelegond | (btw notmyname, the problem with my 'swift-get-nodes' is because I'm kinda of a dumbass, and used the account name instead of AUTH_<hash>) | 18:04 |
| khelegond | Thanks for the help guys! | 18:04 |
| notmyname | glad you figured it out! | 18:04 |
| *** itlinux has joined #openstack-swift | 18:16 | |
| *** e0ne has joined #openstack-swift | 19:07 | |
| *** mikecmpbll has joined #openstack-swift | 19:27 | |
| *** khelegond has quit IRC | 19:33 | |
| *** mvkr has joined #openstack-swift | 20:32 | |
| *** itlinux_ has joined #openstack-swift | 20:36 | |
| *** itlinux has quit IRC | 20:37 | |
| *** itlinux_ has quit IRC | 20:39 | |
| clayg | timburke: I think the canonical implementation of "left hand partner" (_get_partners at the top of reconstructor) also relies on (0 - 1) % 7 == 6 🤷♂️ | 21:40 |
| clayg | I thought that was just how modulo math worked, but IANAM | 21:40 |
| timburke | apparently there's a rather long table at https://en.wikipedia.org/wiki/Modulo_operation that enumerates the positions staked out by various languages | 21:42 |
| timburke | ha! it even includes MUMPS in the list! | 21:42 |
| notmyname | I think I got a vaccine for that when I was a kid. | 21:44 |
| timburke | me too, but then i still wound up working somewhere where i had to use is :-( | 21:45 |
| notmyname | timburke: is there any s3 copy operation that we'll 405 on? (or other s3 verb?) | 21:49 |
| timburke | not that i'm aware of, why? | 21:50 |
| timburke | iirc they implement COPY as a form of PUT | 21:50 |
| notmyname | heard a rumor of a 405 somewhere, and was only checking for the obvious. I'm not aware of anything that would 405, either, but I wanted to check | 21:51 |
| timburke | yeah, https://docs.aws.amazon.com/AmazonS3/latest/API/RESTObjectCOPY.html | 21:52 |
| timburke | oh! SELECT | 21:52 |
| timburke | no, that comes through as a POST... | 21:52 |
| *** openstackgerrit has joined #openstack-swift | 22:01 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: misc test cleanup https://review.openstack.org/631077 | 22:01 |
| *** rcernin has joined #openstack-swift | 22:46 | |
| *** e0ne has quit IRC | 23:24 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Store version id when copying object to archive https://review.openstack.org/437523 | 23:24 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Add X-Backend-Versioning-Mode-Override https://review.openstack.org/437196 | 23:24 |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Modify obj GETorHEAD to support versions https://review.openstack.org/631095 | 23:24 |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Listing of versioned objects https://review.openstack.org/631096 | 23:24 |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Delete versioned objects https://review.openstack.org/631097 | 23:24 |
| timburke | let's just see how those do... | 23:25 |
| *** mikecmpbll has quit IRC | 23:58 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!