| clayg | so a balance of 0.0 is perfect, the "your ring is not balanced you should keep balancing" only fires when balance > 5 | 00:04 |
|---|---|---|
| clayg | ok, so get_balance does an abs | 00:05 |
| clayg | ok, it also goes over all devices and just picks whichever device has the worse balance, so ok that makes sense i suppose | 00:06 |
| clayg | but that makes it seem like there is no upper bound, a device could have 200% more parts than it wants | 00:06 |
| notmyname | can't it? | 00:07 |
| *** nellysmitt has joined #openstack-swift | 00:09 | |
| clayg | yeah i mean we special case weight of zero to MAX_BALANCE (999.0), but if you just set the weight to a really small float you get a super huge balance, this device has 256 parts, and that's ONE MILLIONS % more than it wants) | 00:10 |
| *** nellysmitt has quit IRC | 00:13 | |
| clayg | so, but I'm not really thinking of doing "dispersion" that way, I think the number that makes the most sense is the % of parts who have more replicas at a tier than that tier wants for replicas - so if you have an unsolveable graph (3 servers, 3:1:1, then you get about 60% of the parts with manage to get a replica on every node, then the last 40% end up having to stick everything in one basket | 00:14 |
| notmyname | ya, dispersion is a cluster (or tier) basis. balance is a drive basis | 00:15 |
| notmyname | right? that's what you're thinking? | 00:15 |
| clayg | acctually I think our greedy most hunry device placement is sort of naive in this fashion - if we know we're not going to solve it we could end up with less parts at risk if we were less optimistic when throwing out replicas for a part and conserved the underweight tiers better :\ | 00:15 |
| clayg | notmyname: yeah, i guess i'm wondering if the "dispersion" should follow a similar pattern as balance and return the maximum of the worst dispersed tier or just the total % of fully dispersed parts in the cluster, maybe I'm not sure when there's a difference | 00:18 |
| notmyname | total % fully dispersed makes the most intuitive sense to me. but I suppose that could be different for each tier | 00:19 |
| notmyname | maybe a per tier number | 00:19 |
| *** Masahiro has joined #openstack-swift | 00:20 | |
| notmyname | X% at the region level, Y% across zones, Z% across servers, AA% across drives? | 00:20 |
| notmyname | AA should be 100. and X might often be 0 | 00:20 |
| notmyname | (eg 1 region) | 00:21 |
| notmyname | or 2 regions | 00:21 |
| clayg | well i'm using "max_replicas_at_tier" to decide when a replica is "fully dispersed" so if you have only one region which holds all replicas thats still <= max_replicas for that region - so it's considered 100% dispersed | 00:22 |
| *** Masahiro has quit IRC | 00:24 | |
| notmyname | ah. maybe there's a subtle difference in how you're defining dispersion | 00:26 |
| notmyname | I was thinking "every replica is in a different failure domain at this tier" | 00:26 |
| *** themadcanudist has left #openstack-swift | 00:27 | |
| clayg | notmyname: I don't think that's terribly useful for a lot of topologies - most of the people having this problem are in situations are because they don't have enough failure domains (3 servers for 3 replicas, two regions for 3 replicas, etc) | 00:28 |
| notmyname | isn't that what we need to expose to them? | 00:28 |
| notmyname | or are you saying they already know that | 00:28 |
| clayg | notmyname: yeah I think there's a term problem here - trying to think of the english | 00:31 |
| *** dmsimard is now known as dmsimard_away | 00:59 | |
| *** addnull has joined #openstack-swift | 01:10 | |
| *** fandi_ has quit IRC | 01:11 | |
| *** gyee has quit IRC | 01:12 | |
| *** masonhsiung has joined #openstack-swift | 01:42 | |
| *** mitz_ has quit IRC | 01:48 | |
| *** mitz_ has joined #openstack-swift | 01:49 | |
| openstackgerrit | Yuan Zhou proposed openstack/swift: EC: HTTP PUT/HEAD changes for object size/etag reporting https://review.openstack.org/135488 | 02:05 |
| *** Masahiro has joined #openstack-swift | 02:09 | |
| *** nellysmitt has joined #openstack-swift | 02:10 | |
| *** haomaiwang has joined #openstack-swift | 02:11 | |
| *** Masahiro has quit IRC | 02:13 | |
| *** nellysmitt has quit IRC | 02:14 | |
| *** aswadr has joined #openstack-swift | 02:22 | |
| *** aix has quit IRC | 02:56 | |
| *** bill_az has quit IRC | 03:17 | |
| *** Masahiro has joined #openstack-swift | 03:57 | |
| *** SkyRocknRoll has joined #openstack-swift | 03:58 | |
| *** SkyRocknRoll has joined #openstack-swift | 03:58 | |
| *** Masahiro has quit IRC | 04:01 | |
| *** nellysmitt has joined #openstack-swift | 04:11 | |
| *** nellysmitt has quit IRC | 04:15 | |
| *** ppai has joined #openstack-swift | 04:59 | |
| *** bkopilov has joined #openstack-swift | 05:05 | |
| *** erlon_ has quit IRC | 05:19 | |
| *** Masahiro has joined #openstack-swift | 05:46 | |
| *** Masahiro has quit IRC | 05:51 | |
| *** kopparam has joined #openstack-swift | 05:54 | |
| *** nshaikh has joined #openstack-swift | 06:11 | |
| *** nellysmitt has joined #openstack-swift | 06:11 | |
| *** nellysmitt has quit IRC | 06:16 | |
| *** wer_ has joined #openstack-swift | 06:33 | |
| *** ppai has quit IRC | 06:44 | |
| *** ppai has joined #openstack-swift | 06:57 | |
| *** Masahiro has joined #openstack-swift | 07:01 | |
| *** ppai has quit IRC | 07:02 | |
| *** ttrumm has joined #openstack-swift | 07:04 | |
| *** ttrumm_ has joined #openstack-swift | 07:06 | |
| *** k4n0 has joined #openstack-swift | 07:07 | |
| *** ttrumm has quit IRC | 07:09 | |
| *** ppai has joined #openstack-swift | 07:16 | |
| *** mitz has quit IRC | 07:22 | |
| *** addnull has quit IRC | 07:26 | |
| *** kopparam has quit IRC | 07:30 | |
| *** addnull has joined #openstack-swift | 07:43 | |
| *** nellysmitt has joined #openstack-swift | 07:49 | |
| *** nshaikh has quit IRC | 07:57 | |
| *** EmilienM|afk is now known as EmilienM | 08:02 | |
| *** kopparam has joined #openstack-swift | 08:04 | |
| *** rledisez has joined #openstack-swift | 08:11 | |
| *** Masahiro has quit IRC | 08:12 | |
| *** geaaru has joined #openstack-swift | 08:45 | |
| *** ppai has quit IRC | 08:57 | |
| *** abhirc has quit IRC | 08:57 | |
| *** jordanP has joined #openstack-swift | 09:03 | |
| *** nshaikh has joined #openstack-swift | 09:04 | |
| *** ppai has joined #openstack-swift | 09:09 | |
| *** Masahiro has joined #openstack-swift | 09:13 | |
| *** kopparam has quit IRC | 09:17 | |
| *** Masahiro has quit IRC | 09:17 | |
| *** kopparam has joined #openstack-swift | 09:20 | |
| *** kopparam has quit IRC | 09:53 | |
| *** kopparam has joined #openstack-swift | 09:57 | |
| *** Masahiro has joined #openstack-swift | 10:14 | |
| *** silor has joined #openstack-swift | 10:14 | |
| *** kopparam has quit IRC | 10:16 | |
| *** kopparam_ has joined #openstack-swift | 10:16 | |
| *** Masahiro has quit IRC | 10:18 | |
| *** EmilienM is now known as EmilienM|afk | 10:22 | |
| *** aix has joined #openstack-swift | 10:32 | |
| *** bkopilov has quit IRC | 10:34 | |
| *** masonhsi_ has joined #openstack-swift | 11:00 | |
| *** ttrumm_ has quit IRC | 11:03 | |
| *** masonhsiung has quit IRC | 11:04 | |
| *** addnull has quit IRC | 11:04 | |
| *** addnull has joined #openstack-swift | 11:05 | |
| *** masonhsi_ has quit IRC | 11:05 | |
| *** EmilienM|afk is now known as EmilienM | 11:07 | |
| *** addnull has quit IRC | 11:10 | |
| *** ttrumm has joined #openstack-swift | 11:25 | |
| *** bkopilov has joined #openstack-swift | 11:26 | |
| *** mahatic has joined #openstack-swift | 11:35 | |
| *** haomaiwang has quit IRC | 11:39 | |
| *** nshaikh has quit IRC | 11:40 | |
| *** silor has quit IRC | 11:44 | |
| *** silor has joined #openstack-swift | 11:45 | |
| *** Masahiro has joined #openstack-swift | 12:02 | |
| *** aswadr has quit IRC | 12:06 | |
| *** Masahiro has quit IRC | 12:07 | |
| *** kopparam_ has quit IRC | 12:11 | |
| *** ttrumm has quit IRC | 12:15 | |
| *** infotection has quit IRC | 12:24 | |
| *** annegentle has quit IRC | 12:24 | |
| *** peluse has quit IRC | 12:24 | |
| *** rledisez has quit IRC | 12:24 | |
| *** wer_ has quit IRC | 12:24 | |
| *** hurricanerix has quit IRC | 12:24 | |
| *** wer has quit IRC | 12:24 | |
| *** tacticus has quit IRC | 12:24 | |
| *** zhiyan has quit IRC | 12:24 | |
| *** nottrobin has quit IRC | 12:24 | |
| *** quack_quack_ has quit IRC | 12:24 | |
| *** StevenK has quit IRC | 12:24 | |
| *** vishy has quit IRC | 12:24 | |
| *** ipolyzos has quit IRC | 12:24 | |
| *** zigo has quit IRC | 12:24 | |
| *** CrackerJackMack has quit IRC | 12:24 | |
| *** briancline has quit IRC | 12:24 | |
| *** mattoliverau has quit IRC | 12:24 | |
| *** acoles_away has quit IRC | 12:24 | |
| *** bkopilov has quit IRC | 12:24 | |
| *** notmyname has quit IRC | 12:24 | |
| *** xianghui has quit IRC | 12:24 | |
| *** mlanner has quit IRC | 12:24 | |
| *** ahale_ has quit IRC | 12:24 | |
| *** zul has quit IRC | 12:24 | |
| *** anticw has quit IRC | 12:24 | |
| *** occupant has quit IRC | 12:24 | |
| *** CybergeekDK has quit IRC | 12:24 | |
| *** clds_ has quit IRC | 12:24 | |
| *** eikke has quit IRC | 12:24 | |
| *** jordanP has quit IRC | 12:24 | |
| *** SkyRocknRoll has quit IRC | 12:24 | |
| *** yuanzz has quit IRC | 12:24 | |
| *** JoshNang has quit IRC | 12:24 | |
| *** amandap has quit IRC | 12:24 | |
| *** cppforlife has quit IRC | 12:24 | |
| *** jokke_ has quit IRC | 12:24 | |
| *** infotection has joined #openstack-swift | 12:25 | |
| *** annegentle has joined #openstack-swift | 12:25 | |
| *** peluse has joined #openstack-swift | 12:25 | |
| *** sendak.freenode.net sets mode: +v peluse | 12:25 | |
| *** rledisez has joined #openstack-swift | 12:26 | |
| *** wer_ has joined #openstack-swift | 12:26 | |
| *** hurricanerix has joined #openstack-swift | 12:26 | |
| *** wer has joined #openstack-swift | 12:26 | |
| *** tacticus has joined #openstack-swift | 12:26 | |
| *** zhiyan has joined #openstack-swift | 12:26 | |
| *** nottrobin has joined #openstack-swift | 12:26 | |
| *** quack_quack_ has joined #openstack-swift | 12:26 | |
| *** StevenK has joined #openstack-swift | 12:26 | |
| *** vishy has joined #openstack-swift | 12:26 | |
| *** ipolyzos has joined #openstack-swift | 12:26 | |
| *** zigo has joined #openstack-swift | 12:26 | |
| *** CrackerJackMack has joined #openstack-swift | 12:26 | |
| *** briancline has joined #openstack-swift | 12:26 | |
| *** mattoliverau has joined #openstack-swift | 12:26 | |
| *** acoles_away has joined #openstack-swift | 12:26 | |
| *** sendak.freenode.net sets mode: +v acoles_away | 12:26 | |
| *** kopparam has joined #openstack-swift | 12:26 | |
| *** bkopilov has joined #openstack-swift | 12:26 | |
| *** notmyname has joined #openstack-swift | 12:26 | |
| *** xianghui has joined #openstack-swift | 12:26 | |
| *** sendak.freenode.net sets mode: +v notmyname | 12:26 | |
| *** mlanner has joined #openstack-swift | 12:26 | |
| *** occupant has joined #openstack-swift | 12:26 | |
| *** ahale_ has joined #openstack-swift | 12:26 | |
| *** zul has joined #openstack-swift | 12:26 | |
| *** anticw has joined #openstack-swift | 12:26 | |
| *** CybergeekDK has joined #openstack-swift | 12:26 | |
| *** clds_ has joined #openstack-swift | 12:26 | |
| *** eikke has joined #openstack-swift | 12:26 | |
| *** jordanP has joined #openstack-swift | 12:28 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:28 | |
| *** yuanzz has joined #openstack-swift | 12:28 | |
| *** JoshNang has joined #openstack-swift | 12:28 | |
| *** amandap has joined #openstack-swift | 12:28 | |
| *** cppforlife has joined #openstack-swift | 12:28 | |
| *** jokke_ has joined #openstack-swift | 12:28 | |
| *** kopparam has quit IRC | 12:43 | |
| *** bill_az has joined #openstack-swift | 12:57 | |
| *** Masahiro has joined #openstack-swift | 13:03 | |
| *** fandi has joined #openstack-swift | 13:06 | |
| *** Masahiro has quit IRC | 13:07 | |
| *** ppai has quit IRC | 13:15 | |
| *** addnull has joined #openstack-swift | 13:15 | |
| *** addnull has quit IRC | 13:29 | |
| *** tab_ has joined #openstack-swift | 13:35 | |
| tab_ | is there any example/case scenario step-by step of setting multi zone and multiregion swift cluster? | 13:36 |
| *** bkopilov has quit IRC | 13:43 | |
| *** tdasilva has joined #openstack-swift | 13:47 | |
| *** addnull has joined #openstack-swift | 13:49 | |
| *** addnull has quit IRC | 13:59 | |
| *** exploreshaifali has joined #openstack-swift | 14:15 | |
| *** addnull has joined #openstack-swift | 14:25 | |
| *** addnull has quit IRC | 14:26 | |
| mahatic | torgomatic, good morning! around? | 14:31 |
| mahatic | torgomatic, too early I guess :) | 14:50 |
| *** Masahiro has joined #openstack-swift | 14:52 | |
| *** Masahiro has quit IRC | 14:56 | |
| openstackgerrit | Jiri Suchomel proposed openstack/python-swiftclient: Print info message about incorrect --totals usage when neither -l nor --lh is provided. Added test coverage for --totals. https://review.openstack.org/130463 | 14:57 |
| *** exploreshaifali has quit IRC | 14:59 | |
| *** dmsimard_away is now known as dmsimard | 15:04 | |
| *** jest has joined #openstack-swift | 15:07 | |
| *** k4n0 has quit IRC | 15:25 | |
| *** nellysmitt has quit IRC | 15:28 | |
| *** acoles_away is now known as acoles | 15:30 | |
| *** exploreshaifali has joined #openstack-swift | 15:39 | |
| *** tdasilva has quit IRC | 15:41 | |
| *** nellysmitt has joined #openstack-swift | 15:42 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Update container on fast-POST https://review.openstack.org/135380 | 15:53 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Make ssync compatible with fast-post meta files https://review.openstack.org/138498 | 15:53 |
| *** exploreshaifali has quit IRC | 15:58 | |
| *** tdasilva has joined #openstack-swift | 16:01 | |
| openstackgerrit | Mahati proposed openstack/swift: Implement OPTIONS verb for storage nodes. https://review.openstack.org/140103 | 16:04 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Update container on fast-POST https://review.openstack.org/135380 | 16:06 |
| *** masonhsiung has joined #openstack-swift | 16:11 | |
| acoles | tdasilva: hi, you here? | 16:23 |
| *** Masahiro has joined #openstack-swift | 16:41 | |
| *** rdaly2 has joined #openstack-swift | 16:42 | |
| *** Masahiro has quit IRC | 16:45 | |
| acoles | tdasilva: ok, sorry to miss you, i looked at https://review.openstack.org/#/c/123765/ and it only covers slo, but like you said it looks like it did cover dlo :S I'm not sure about the extra HEAD that it introduces. | 16:49 |
| *** lpabon has joined #openstack-swift | 16:50 | |
| acoles | tdasilva: anyway thats as far as i got - have a great holiday! i'm ooo until the new year. | 16:50 |
| *** nellysmitt has quit IRC | 16:54 | |
| *** geaaru has quit IRC | 16:55 | |
| *** exploreshaifali has joined #openstack-swift | 17:01 | |
| *** hugokuo has quit IRC | 17:08 | |
| *** ttrumm has joined #openstack-swift | 17:10 | |
| *** nellysmitt has joined #openstack-swift | 17:12 | |
| *** bkopilov has joined #openstack-swift | 17:14 | |
| jest | hi there. is there any example reference configuration for multi-region, multi-zone cluster for Swift? | 17:14 |
| *** acoles is now known as acoles_away | 17:15 | |
| *** hugokuo has joined #openstack-swift | 17:15 | |
| *** ttrumm has quit IRC | 17:17 | |
| *** nellysmitt has quit IRC | 17:25 | |
| *** lpabon has quit IRC | 17:28 | |
| *** rledisez has quit IRC | 17:31 | |
| *** openstackgerrit has quit IRC | 17:34 | |
| *** openstackgerrit has joined #openstack-swift | 17:35 | |
| *** fandi has quit IRC | 17:39 | |
| *** jordanP has quit IRC | 17:42 | |
| *** fandi has joined #openstack-swift | 17:49 | |
| notmyname | good morning | 17:56 |
| *** masonhsiung has quit IRC | 18:06 | |
| *** masonhsiung has joined #openstack-swift | 18:06 | |
| *** silor has quit IRC | 18:07 | |
| *** masonhsiung has quit IRC | 18:10 | |
| *** Masahiro has joined #openstack-swift | 18:29 | |
| *** Masahiro has quit IRC | 18:34 | |
| mahatic | notmyname, good morning. Is the opw meeting on tomorrow? | 18:35 |
| notmyname | mahatic: no. not next week either. I'll be traveling. however, I will be on from time to time during the next 1.5 weeks | 18:39 |
| *** exploreshaifali has quit IRC | 18:43 | |
| *** zaitcev has joined #openstack-swift | 19:21 | |
| *** ChanServ sets mode: +v zaitcev | 19:21 | |
| *** nellysmitt has joined #openstack-swift | 19:26 | |
| *** nellysmitt has quit IRC | 19:30 | |
| *** exploreshaifali has joined #openstack-swift | 19:32 | |
| tdasilva | acoles_away: hi, sorry to have missed you earlier. I saw your comments on the review and agree everthing. I'll look to implement the same changes in the obj. versioning middleware | 19:35 |
| *** mahatic has quit IRC | 19:39 | |
| *** fifieldt has quit IRC | 20:03 | |
| notmyname | remember that there is not a swift team meeting tomorrow or next week | 20:05 |
| tdasilva | notmyname: thanks! enjoy the break :-) | 20:05 |
| notmyname | you too | 20:05 |
| *** fifieldt has joined #openstack-swift | 20:16 | |
| *** Masahiro has joined #openstack-swift | 20:18 | |
| *** abhirc has joined #openstack-swift | 20:21 | |
| *** Masahiro has quit IRC | 20:22 | |
| *** bill_az has quit IRC | 20:40 | |
| *** mahatic has joined #openstack-swift | 20:46 | |
| *** Nadeem_ has joined #openstack-swift | 20:59 | |
| *** EmilienM is now known as EmilienM|afk | 21:04 | |
| mattoliverau | morning all on this christmas eve day | 21:25 |
| notmyname | hi! | 21:25 |
| notmyname | anyone running stud as an ssl/tls terminator in front of swift? | 21:26 |
| *** nellysmitt has joined #openstack-swift | 21:27 | |
| *** nellysmitt has quit IRC | 21:32 | |
| *** EmilienM|afk is now known as EmilienM | 21:48 | |
| *** Masahiro has joined #openstack-swift | 22:07 | |
| *** Masahiro has quit IRC | 22:11 | |
| *** esmute has quit IRC | 22:17 | |
| *** esmute has joined #openstack-swift | 22:46 | |
| *** jest has quit IRC | 22:54 | |
| *** dmsimard is now known as dmsimard_away | 23:20 | |
| *** nellysmitt has joined #openstack-swift | 23:28 | |
| *** nellysmitt has quit IRC | 23:32 | |
| *** exploreshaifali has quit IRC | 23:46 | |
| *** miqui_ has quit IRC | 23:46 | |
| *** Masahiro has joined #openstack-swift | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!