| *** dhellmann has joined #openstack-swift | 00:07 | |
| *** gyee has quit IRC | 00:55 | |
| *** SkyRocknRoll has joined #openstack-swift | 01:42 | |
| openstackgerrit | liuyamin proposed openstack/swift master: Remove setup.py check from pep8 job https://review.openstack.org/616394 | 02:27 |
|---|---|---|
| *** fungi has quit IRC | 03:06 | |
| *** fungi has joined #openstack-swift | 03:09 | |
| *** fungi has quit IRC | 03:10 | |
| *** fungi has joined #openstack-swift | 03:40 | |
| *** fungi has quit IRC | 03:41 | |
| *** fungi has joined #openstack-swift | 03:45 | |
| *** SkyRocknRoll has quit IRC | 07:17 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:21 | |
| *** SkyRocknRoll has quit IRC | 07:24 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:25 | |
| *** pcaruana has joined #openstack-swift | 07:34 | |
| *** SkyRocknRoll has quit IRC | 07:44 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:57 | |
| *** e0ne has joined #openstack-swift | 09:50 | |
| *** mvkr has quit IRC | 09:58 | |
| *** mvkr has joined #openstack-swift | 10:26 | |
| *** admin6_ has joined #openstack-swift | 12:14 | |
| *** admin6 has quit IRC | 12:16 | |
| *** admin6_ is now known as admin6 | 12:16 | |
| *** admin6 has quit IRC | 12:24 | |
| cwright | notmyname: I'm just now seeing the swift-exporter announcement, thank you very much! | 12:39 |
| *** ccamacho has quit IRC | 12:53 | |
| *** e0ne has quit IRC | 13:03 | |
| *** SkyRocknRoll has quit IRC | 13:16 | |
| *** ccamacho has joined #openstack-swift | 13:18 | |
| *** e0ne has joined #openstack-swift | 13:37 | |
| *** ccamacho has quit IRC | 13:52 | |
| *** ccamacho has joined #openstack-swift | 13:53 | |
| *** ccamacho has quit IRC | 14:02 | |
| *** kota_ has quit IRC | 14:02 | |
| *** irclogbot_2 has quit IRC | 14:02 | |
| *** kukacz has quit IRC | 14:02 | |
| *** tdasilva has quit IRC | 14:02 | |
| *** mahatic has quit IRC | 14:02 | |
| *** tonyb has quit IRC | 14:02 | |
| *** mattoliverau has quit IRC | 14:02 | |
| *** kukacz has joined #openstack-swift | 14:04 | |
| *** kota_ has joined #openstack-swift | 14:04 | |
| *** ChanServ sets mode: +v kota_ | 14:04 | |
| *** mahatic has joined #openstack-swift | 14:04 | |
| *** ChanServ sets mode: +v mahatic | 14:04 | |
| *** tonyb has joined #openstack-swift | 14:07 | |
| *** mvkr has quit IRC | 15:04 | |
| *** ccamacho has joined #openstack-swift | 15:09 | |
| *** tdasilva has joined #openstack-swift | 15:11 | |
| *** ChanServ sets mode: +v tdasilva | 15:11 | |
| *** mvkr has joined #openstack-swift | 15:20 | |
| *** admin6 has joined #openstack-swift | 15:48 | |
| admin6 | Hi there, I have a small question about the "unique-as-possible" distribution of data. I currently run a 9+3 erasure coding ring over 6 servers. Each server/node currently is declared in a different zone. I’m adding two new servers but I don’t want to add new zones to the ring. What is the best solution to add these new servers : add them only in one zone each, but this will unbalance the zones with zone 1 and | 16:06 |
| admin6 | having twice the storage of zone 3 to 6? Is that a problem ? Or is there a better solution to integrate only 2 new servers for the data distribution ? | 16:06 |
| DHE | you can add and remove zones freely. there isn't really a "process" for making zones. you just declare the server is a member of zone X | 16:10 |
| *** pcaruana has quit IRC | 16:14 | |
| *** mvkr has quit IRC | 16:16 | |
| *** gyee has joined #openstack-swift | 16:18 | |
| tdasilva | admin6: do you already have a lot of data in your cluster? | 16:23 |
| tdasilva | the reason I'm asking is because my first thought is to also just change the rings to reduce the number of zones you have, but i'm not 100% if that will cause a whole lot of data movement | 16:24 |
| tdasilva | but you are correct in trying to avoid having zones being unbalanced | 16:25 |
| tdasilva | the 'unique-as-possible' calculations will already take into effect the nodes (i.e., IP address) being a failure domain, so there's no point in "making" 1 zone per node | 16:27 |
| tdasilva | in fact you could be losing the opportunity to declare a rack as a failure domain for example | 16:28 |
| *** mvkr has joined #openstack-swift | 16:28 | |
| DHE | "as uniquely as possible" just means trying to spread as best it can at the highest level (region), then the next level (zone), then the next (node/host). having 1 host per zone doesn't really accomplish anything. | 16:28 |
| DHE | unless they're already in discrete failure domains (eg: discrete racks, individual PDUs and switches) and you're planning for future growth | 16:30 |
| tdasilva | exactly | 16:35 |
| admin6 | Yes, the servers are one different rack, but I have not enough budget to buy 6 new servers at at the same time, only two… So I’m trying to find the best solution. The other option I considered is to reorganize and reduce to only 4 zones, gradually moving server 5 and 6 to zone 3 and 4 and adding the new servers to zone 1 and 2. | 16:45 |
| admin6 | but I will have server over multi-zones for a time and that mean I could have more than 3 fragments on a single node, making it a point of failure for some objects. | 16:47 |
| *** e0ne has quit IRC | 16:48 | |
| admin6 | unless if the "unique-as-possible" distribution is ables to recognize that it is the same node (ip/port) even if the disks are in different zones. but I doubt this is the case. | 16:49 |
| admin6 | tdasilva: yes, the disks are between 80 and 85% full on each server and I’m storing about 600TB in this ring, that’s why I’m adding new servers. | 16:57 |
| admin6 | tdasilva: thanks for your answer, I need to leave but I’ll reconnect in a couple of hours :-) | 17:03 |
| *** admin6 has left #openstack-swift | 17:03 | |
| notmyname | good morning | 17:28 |
| *** ccamacho has quit IRC | 17:32 | |
| *** mvkr has quit IRC | 17:35 | |
| *** irclogbot_2 has joined #openstack-swift | 18:02 | |
| *** pcaruana has joined #openstack-swift | 18:10 | |
| *** SkyRocknRoll has joined #openstack-swift | 18:17 | |
| openstackgerrit | Merged openstack/swift master: Remove setup.py check from pep8 job https://review.openstack.org/616394 | 18:22 |
| *** SkyRocknRoll has quit IRC | 18:29 | |
| *** e0ne has joined #openstack-swift | 20:19 | |
| *** e0ne has quit IRC | 20:22 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Add basic support for ?versions bucket listings https://review.openstack.org/575838 | 21:14 |
| *** mattoliverau has joined #openstack-swift | 22:07 | |
| *** ChanServ sets mode: +v mattoliverau | 22:07 | |
| mattoliverau | morning, damn irc client/bouncer reconnected, and so needed to reauthise so couldn't auto join the channel.. I should really look into auto reauth. I wonder if this irc client can do that. | 22:13 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!