| *** chsc has quit IRC | 00:03 | |
| *** vinsh has joined #openstack-swift | 00:06 | |
| *** jamielennox is now known as jamielennox|away | 00:25 | |
| *** zhurong has joined #openstack-swift | 00:30 | |
| *** jamielennox|away is now known as jamielennox | 00:31 | |
| *** vint_bra has joined #openstack-swift | 00:53 | |
| kota_ | good morning | 00:55 |
|---|---|---|
| tone_z | Good morning! | 01:02 |
| tone_z | mattoliverau: morning! | 01:02 |
| tone_z | kota_: morning! | 01:02 |
| *** vinsh has quit IRC | 01:03 | |
| mattoliverau | kota_, tone_z: morning | 01:03 |
| kota_ | tone_z, mattoliverau: \o/ | 01:04 |
| *** vint_bra has quit IRC | 01:10 | |
| kota_ | mattoliverau: i did NOT see the code in patch 448480 yet, how do you think of the bug 167500? | 01:26 |
| openstack | bug 167500 in Inkscape "Rulers don't align with document " [High,Fix released] https://launchpad.net/bugs/167500 - Assigned to Diederik van Lierop (mail-diedenrezi) | 01:26 |
| patchbot | https://review.openstack.org/#/c/448480/ - swift - Container drive error results double space usage o... | 01:26 |
| kota_ | IDK the patch 448480 works for the case that 2 handoff nodes attempt to place the replica db to where.... | 01:27 |
| patchbot | https://review.openstack.org/#/c/448480/ - swift - Container drive error results double space usage o... | 01:27 |
| kota_ | 3 other nodes except the handoff local? or 3 nodes including itself | 01:28 |
| kota_ | the latter one could work well but the prior one may be terrible because the handoff nodes push their local each other (i.e. racing), if i understand the problem correctly | 01:29 |
| * kota_ means 3 replicas case | 01:29 | |
| *** sams-gleb has joined #openstack-swift | 01:38 | |
| mattoliverau | kota_: I'l need to test but this version, just keeps track of having to send to len(repl_nodes) nodes, on each node it decrements and if one comes back as a DirveNotMounted it'll simply icrement the counter, and thus fall into more_nodes. It's stopping the scan to find the handoff node, so will alwasy try and hit up the same handoff nodes. So it should scale to any number of replicas. But I'll check on my SAIO with | 01:40 |
| mattoliverau | different replicas to be sure. | 01:40 |
| mattoliverau | if it hits itself as a hand off node, then like in the reconstructor it'll just continue (after decrementing) thereby skipping as he's already a handoff node. | 01:41 |
| kota_ | mattoliverau: thx for sharing the status, that's helpful to understand | 01:41 |
| mattoliverau | I just we should confirm that it wont then delete off itself. | 01:42 |
| kota_ | sounds reasenable (just continue after decrementing) | 01:42 |
| mattoliverau | otherwise we'd loose redundency | 01:42 |
| kota_ | wondering the latter handoff attempts to try to push over prior handoff? | 01:42 |
| kota_ | e.g. | 01:42 |
| *** sams-gleb has quit IRC | 01:43 | |
| kota_ | p0, p1, p2, h1, h2, h3... and if h3 has a replica as a handoff | 01:43 |
| kota_ | for some reasons (e.g. rebalance? network failure?), anyway | 01:44 |
| kota_ | and then p0 is now failing with the device to store so p0 will raise device not mounted | 01:45 |
| mattoliverau | so h3 will try and push to p0, p1, p2, h1, h2 in that order. | 01:45 |
| kota_ | and then, h3 tries to p1, p2, h1? | 01:45 |
| mattoliverau | but yeah only on DriveNotMounted any other failure and it wont push it, just error. Like we currently do. | 01:45 |
| kota_ | because the h3 is unable to find itself in the order. | 01:45 |
| kota_ | untill h3 comes | 01:45 |
| mattoliverau | kota_: oh yeah, I guess it would push to a handoff node in that case. | 01:46 |
| mattoliverau | Maybe I'll need to go and add a test to see :) | 01:46 |
| mattoliverau | question what behaviour do we want. in that case, push to a handoff node? or hey I'm a handoff node so I'm already good. | 01:48 |
| *** JimCheung has quit IRC | 01:48 | |
| mattoliverau | except in the latter case, what happens its been a rebalance to remove the drive | 01:48 |
| *** JimCheung has joined #openstack-swift | 01:49 | |
| kota_ | AFAIK, the object-replicator attempts to push the local ONLY to the primaries | 01:49 |
| kota_ | exactly, it may be a tradeoff (racing to create handoffs vs it may decrease durability if we leave the device failures for a long term) | 01:50 |
| kota_ | or the mid solution may be trying to push the local to fill 3 replicas including his local | 01:51 |
| kota_ | so attempt_left will start from... | 01:52 |
| kota_ | ah.. it needs to handle the primary or the handoff separately, in the handoff, # of primaries - # of primary failures - 1 (the handoff) that will be attempt_left | 01:53 |
| *** JimCheung has quit IRC | 01:53 | |
| kota_ | maybe? | 01:53 |
| kota_ | idk, and not sure | 01:53 |
| kota_ | ah, no | 01:54 |
| kota_ | the consition is not true | 01:54 |
| kota_ | so, anyway, try primaries at first. | 01:55 |
| kota_ | and then... if 1 failure found, and the replicator is in the handoff, it sounds fine, no more handoff you need to try to push. | 01:55 |
| kota_ | if 2 failrues found, the replicator need to push 1 another handoff to keep the durability | 01:56 |
| mattoliverau | in your simulation before: p1, p2, p3, h1, h2, h3 and h3 has the handoff, we can make it so it says cool we need 3, I can be one.. but if p2 is down too, then it'll place it on p1, h1, h3 (itself).. but h1 and would place it on p1, h1 (itself), h2... which at least would be the same as now if p1 ran. | 01:56 |
| kota_ | how do we express the condition | 01:56 |
| mattoliverau | So that I'm saying it, we can still get into a situation where there is an extra handoff and it wont remove it because it doesn't know about the others.. but maybe that's ok? | 01:57 |
| kota_ | mattoliverau: true | 01:57 |
| kota_ | sounds terrible, isn't it? | 01:57 |
| mattoliverau | hmm, it does kinda.. well until a rebalance or the primary is fixed... which would always be the problem in any case, except in this case there might be an extra handoff out there somewhere | 01:58 |
| mattoliverau | tho thats better then now.. where we place it everywhere :P | 01:59 |
| kota_ | mattoliverau: exactly, current one in the master is the worst :P | 01:59 |
| mattoliverau | well in the current patch, h3 would pus to p1, h1, h2 and then delete itself.. which would mean we still have durability (as far as its concerned) and now using up any extra space.. but there could be some more movement. However h1, and h2 wont ever move it again until the primaries are fixed or a rebalance happens.. so maybe it is better | 02:01 |
| mattoliverau | *now not using up extra space | 02:02 |
| mattoliverau | kota_: sorry, my typing seems to be bad, which is making my english bad :P | 02:02 |
| mattoliverau | I'll try and read my text before I press send :) | 02:03 |
| kota_ | mattoliverau: ok, thanks for the info. sorry, I have to leave here for a while for a meeting | 02:04 |
| mattoliverau | kota_: no probs :) | 02:04 |
| mattoliverau | I'll do some playing with the patch | 02:05 |
| mattoliverau | kota_: I know your going to a meeting, but what I said before was slightly wrong, in the current patch h3 after pushing to p1, h1, h3 wouldn't delete itself because it would have got a failure from p2 and p3. So the data would also just stick around. If we want to clean itself, then we'd need to look for num replica successes in responses.(line 589) but thats only if we want to go that approach :) | 02:14 |
| *** sams-gleb has joined #openstack-swift | 03:40 | |
| *** klrmn has quit IRC | 03:44 | |
| *** sams-gleb has quit IRC | 03:45 | |
| *** chsc has joined #openstack-swift | 03:57 | |
| *** JimCheung has joined #openstack-swift | 04:03 | |
| *** JimCheung has quit IRC | 04:07 | |
| *** zhurong has quit IRC | 04:09 | |
| *** kei_yama has quit IRC | 04:10 | |
| *** chsc has quit IRC | 04:13 | |
| *** gkadam has joined #openstack-swift | 04:17 | |
| *** zhurong has joined #openstack-swift | 04:30 | |
| *** psachin has joined #openstack-swift | 04:42 | |
| *** kei_yama has joined #openstack-swift | 04:46 | |
| *** _JZ_ has quit IRC | 04:53 | |
| *** ChubYann has quit IRC | 05:18 | |
| *** SkyRocknRoll has quit IRC | 05:34 | |
| *** zhurong has quit IRC | 05:41 | |
| *** sams-gleb has joined #openstack-swift | 05:43 | |
| *** rcernin has joined #openstack-swift | 05:44 | |
| *** sams-gleb has quit IRC | 05:47 | |
| *** jaosorior has joined #openstack-swift | 05:56 | |
| *** zhurong has joined #openstack-swift | 06:05 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift master: sharding - Fix container server put/delete piv https://review.openstack.org/451197 | 06:12 |
| *** jojden has joined #openstack-swift | 06:15 | |
| *** chmouel has left #openstack-swift | 06:33 | |
| *** hseipp has joined #openstack-swift | 06:38 | |
| PavelK | kota_: mattoliverau: with patch 448480 - when h3 finds a replica that belongs to p1, p2, p3 and handoffs h1, h2, h3 then it pushes it to primaries and if necessary to handoffs in the ordes h1, h2. It doesn't delete the replica until it is pushed to all primary nodes (as object replicator does). I think it doesn't matter. If you reballance then we expect to have all primary drives health and handoffs | 06:39 |
| patchbot | https://review.openstack.org/#/c/448480/ - swift - Container drive error results double space usage o... | 06:39 |
| PavelK | are cleared shortly. | 06:39 |
| *** sams-gleb has joined #openstack-swift | 06:47 | |
| *** d0ugal has joined #openstack-swift | 06:52 | |
| *** d0ugal has quit IRC | 06:54 | |
| *** d0ugal has joined #openstack-swift | 06:58 | |
| *** d0ugal has quit IRC | 06:58 | |
| *** d0ugal has joined #openstack-swift | 06:58 | |
| *** JimCheung has joined #openstack-swift | 07:02 | |
| *** JimCheung has quit IRC | 07:06 | |
| *** tesseract has joined #openstack-swift | 07:08 | |
| *** geaaru has joined #openstack-swift | 07:19 | |
| *** d0ugal has quit IRC | 07:22 | |
| *** d0ugal has joined #openstack-swift | 07:26 | |
| *** d0ugal has quit IRC | 07:26 | |
| *** d0ugal has joined #openstack-swift | 07:26 | |
| *** pcaruana has joined #openstack-swift | 07:27 | |
| -openstackstatus- NOTICE: Jobs in gate are failing with POST_FAILURE. Infra roots are investigating | 07:45 | |
| *** ChanServ changes topic to "Jobs in gate are failing with POST_FAILURE. Infra roots are investigating" | 07:45 | |
| *** rcernin has quit IRC | 07:54 | |
| *** tesseract has quit IRC | 07:54 | |
| *** tesseract has joined #openstack-swift | 07:55 | |
| *** rcernin has joined #openstack-swift | 07:56 | |
| -openstackstatus- NOTICE: logs.openstack.org has corrupted disks, it's being repaired. Please avoid rechecking until this is fixed | 08:25 | |
| *** ChanServ changes topic to "logs.openstack.org has corrupted disks, it's being repaired. Please avoid rechecking until this is fixed" | 08:25 | |
| *** cbartz has joined #openstack-swift | 08:31 | |
| mattoliverau | PavelK: I'm not actually at work (ie I'm suppose to be with the family) but i'm not disagreeing, I'm not saynig the bahaviour is wrong. Which is why I didn't score. I'm not against keeping the extra around. Just wanted to make sure the discusson was had as a community, and so made a probe test. Which ever way we decide we can always change the probe test to suit and then at least we'd have a test to keep it correct ;) | 09:18 |
| PavelK | mattoliverau: I understand that you do not hate it :-). I agree that it needs to be discussed more. | 09:25 |
| *** peterlisak has quit IRC | 09:28 | |
| *** onovy has quit IRC | 09:28 | |
| *** psachin has quit IRC | 09:36 | |
| *** peterlisak has joined #openstack-swift | 09:39 | |
| openstackgerrit | Romain LE DISEZ proposed openstack/swift master: Fix encoding issue in ssync_sender.send_put() https://review.openstack.org/452112 | 09:39 |
| rledisez | hello all, can someone have a lot at ^ please? | 09:40 |
| rledisez | it fixes https://bugs.launchpad.net/swift/+bug/1678018 which seems like a critical issue in reconstructor | 09:40 |
| openstack | Launchpad bug 1678018 in OpenStack Object Storage (swift) "ssync_sender encoding issue" [Undecided,New] | 09:40 |
| *** onovy has joined #openstack-swift | 09:43 | |
| *** psachin has joined #openstack-swift | 09:54 | |
| *** sputnik13 has joined #openstack-swift | 10:03 | |
| *** jarbod__ has joined #openstack-swift | 10:06 | |
| openstackgerrit | Gábor Antal proposed openstack/swift master: Use more specific asserts in test/unit/container https://review.openstack.org/342808 | 10:10 |
| *** sams-gleb has quit IRC | 10:12 | |
| *** sams-gleb has joined #openstack-swift | 10:12 | |
| *** sams-gleb has quit IRC | 10:16 | |
| *** zhurong has quit IRC | 10:18 | |
| *** zhurong has joined #openstack-swift | 10:41 | |
| *** psachin has quit IRC | 10:50 | |
| *** psachin has joined #openstack-swift | 10:52 | |
| *** hseipp has quit IRC | 11:02 | |
| *** JimCheung has joined #openstack-swift | 11:03 | |
| *** JimCheung has quit IRC | 11:08 | |
| *** sams-gleb has joined #openstack-swift | 11:14 | |
| *** hseipp has joined #openstack-swift | 11:26 | |
| *** vint_bra has joined #openstack-swift | 11:32 | |
| *** kei_yama has quit IRC | 11:51 | |
| *** zhurong has quit IRC | 11:59 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:05 | |
| *** SkyRocknRoll has quit IRC | 12:06 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:07 | |
| *** vint_bra has quit IRC | 12:21 | |
| *** SkyRocknRoll has quit IRC | 12:22 | |
| *** catintheroof has joined #openstack-swift | 12:36 | |
| *** chlong has joined #openstack-swift | 12:40 | |
| *** hseipp has quit IRC | 12:41 | |
| *** hseipp has joined #openstack-swift | 12:41 | |
| *** ouchkernel has quit IRC | 12:51 | |
| *** HW-Peter has joined #openstack-swift | 12:55 | |
| *** HW-Peter has quit IRC | 12:56 | |
| *** HW-Peter has joined #openstack-swift | 12:56 | |
| *** gkadam has quit IRC | 12:57 | |
| *** PavelK has quit IRC | 13:05 | |
| jrichli | rledisez: I can take a look. I can't do it in the first half of my day, but can start on it in the afternoon. | 13:07 |
| *** PavelK has joined #openstack-swift | 13:25 | |
| *** vint_bra has joined #openstack-swift | 13:26 | |
| *** adu has joined #openstack-swift | 13:27 | |
| *** mariusv_ has quit IRC | 13:34 | |
| *** mariusv has joined #openstack-swift | 13:35 | |
| *** mariusv has joined #openstack-swift | 13:35 | |
| *** PavelK has quit IRC | 13:46 | |
| rledisez | jrichli: thx :) | 13:51 |
| *** amoralej is now known as amoralej|lunch | 13:54 | |
| *** Dinesh_Bhor has quit IRC | 13:55 | |
| *** JimCheung has joined #openstack-swift | 14:03 | |
| *** geaaru has quit IRC | 14:04 | |
| *** ouchkernel has joined #openstack-swift | 14:07 | |
| *** JimCheung has quit IRC | 14:08 | |
| *** jamielennox is now known as jamielennox|away | 14:16 | |
| *** geaaru has joined #openstack-swift | 14:17 | |
| *** jroll has quit IRC | 14:32 | |
| *** jroll has joined #openstack-swift | 14:34 | |
| *** amoralej|lunch is now known as amoralej | 14:36 | |
| *** vinsh has joined #openstack-swift | 14:52 | |
| *** rcernin has quit IRC | 15:08 | |
| *** chsc has joined #openstack-swift | 15:11 | |
| *** chsc has joined #openstack-swift | 15:11 | |
| *** psachin has quit IRC | 15:13 | |
| *** klrmn has joined #openstack-swift | 15:13 | |
| *** chsc has quit IRC | 15:27 | |
| *** chsc has joined #openstack-swift | 15:27 | |
| *** chsc has joined #openstack-swift | 15:27 | |
| *** jaosorior has quit IRC | 15:28 | |
| *** jojden has quit IRC | 15:30 | |
| *** klrmn has quit IRC | 15:35 | |
| *** pcaruana has quit IRC | 15:46 | |
| *** jordanP has quit IRC | 15:49 | |
| *** _JZ_ has joined #openstack-swift | 16:06 | |
| *** sams-gleb has quit IRC | 16:06 | |
| *** hseipp has quit IRC | 16:20 | |
| *** JimCheung has joined #openstack-swift | 16:24 | |
| *** sams-gleb has joined #openstack-swift | 16:28 | |
| *** geaaru has quit IRC | 16:47 | |
| *** oshritf has joined #openstack-swift | 16:53 | |
| *** cbartz has quit IRC | 16:54 | |
| *** klrmn has joined #openstack-swift | 16:56 | |
| *** gyee has joined #openstack-swift | 16:59 | |
| *** MVenesio has joined #openstack-swift | 17:18 | |
| MVenesio | Hi Guys, i’ve installed a swift cloud to a client with 1 region, 3 zones, 3 servers, and 3 copies configured using replication. The particularity of this servers, is that one of them has less disks and less space to store partitions than the others. While i’m testing the data placement and replication, i’m figuring out that servers #1 and #3 are storing the 3 replicas of all objects as is expected, since those two servers hav | 17:19 |
| MVenesio | e more disks and space. The question is, when the system will start to place data in the server #2 also ? Is it related with the amount of space or partitions used, or something else ? | 17:19 |
| notmyname | MVenesio: start with the dispersion and overload sections on https://docs.openstack.org/developer/swift/overview_ring.html | 17:20 |
| MVenesio | @notmyname thanks man | 17:21 |
| *** bikmak has quit IRC | 17:22 | |
| *** klrmn has quit IRC | 17:22 | |
| *** klrmn has joined #openstack-swift | 17:24 | |
| *** chsc has quit IRC | 17:29 | |
| *** oshritf has quit IRC | 17:31 | |
| *** ChubYann has joined #openstack-swift | 17:35 | |
| clayg | hello all | 17:41 |
| notmyname | clayg: hello | 17:59 |
| *** chsc has joined #openstack-swift | 18:19 | |
| *** chsc has joined #openstack-swift | 18:19 | |
| clayg | notmyname: https://wiki.openstack.org/wiki/Swift/PriorityReviews is out of date | 18:19 |
| notmyname | indeed | 18:20 |
| *** tesseract has quit IRC | 18:21 | |
| notmyname | clayg: reload. not helpful, but more correct | 18:24 |
| clayg | lol | 18:25 |
| clayg | yeah much better | 18:25 |
| *** klrmn has quit IRC | 18:53 | |
| *** amoralej is now known as amoralej|off | 18:54 | |
| *** HW-Peter has quit IRC | 19:06 | |
| *** MVenesio has quit IRC | 19:10 | |
| *** hseipp has joined #openstack-swift | 19:11 | |
| *** klrmn has joined #openstack-swift | 19:33 | |
| *** hseipp has quit IRC | 19:45 | |
| *** klrmn has quit IRC | 19:45 | |
| *** klrmn has joined #openstack-swift | 19:55 | |
| -openstackstatus- NOTICE: lists.openstack.org will be offline from 20:00 to 23:00 UTC for planned upgrade maintenance | 19:59 | |
| *** klrmn has quit IRC | 20:05 | |
| *** stradling has joined #openstack-swift | 20:10 | |
| *** stradling has quit IRC | 20:42 | |
| *** klrmn has joined #openstack-swift | 21:16 | |
| *** mvk has quit IRC | 21:18 | |
| *** vinsh has quit IRC | 21:36 | |
| *** klrmn has quit IRC | 21:40 | |
| *** klrmn has joined #openstack-swift | 21:40 | |
| -openstackstatus- NOTICE: The upgrade maintenance for lists.openstack.org has been completed and it is back online. | 21:51 | |
| *** vint_bra has quit IRC | 21:55 | |
| *** sams-gleb has quit IRC | 22:20 | |
| *** sams-gleb has joined #openstack-swift | 22:20 | |
| *** sams-gleb has quit IRC | 22:21 | |
| *** klrmn has quit IRC | 22:24 | |
| *** catintheroof has quit IRC | 22:36 | |
| *** _JZ_ has quit IRC | 23:04 | |
| *** sams-gleb has joined #openstack-swift | 23:21 | |
| *** sams-gleb has quit IRC | 23:27 | |
| *** klrmn has joined #openstack-swift | 23:37 | |
| *** chsc has quit IRC | 23:51 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!