| *** cleverdevil has quit IRC | 00:04 | |
| *** puranamr has joined #akanda | 16:04 | |
| *** cleverdevil has joined #akanda | 16:29 | |
| openstackgerrit | Rama Puranam proposed stackforge/akanda: Implements: blueprint vm-reduce-footprint https://review.openstack.org/207559 | 17:44 |
|---|---|---|
| sarob | puranamr: welcome new contributor! | 17:58 |
| *** puranamr has quit IRC | 17:59 | |
| adam_g | sarob, did you just cut liberty tags or something? | 18:40 |
| adam_g | http://logs.openstack.org/76/198176/4/check/gate-akanda-rug-python27/c64d412/console.html | 18:41 |
| adam_g | sarob, i think you need to also push 7.0.0.0b1 | 18:50 |
| adam_g | and we can push through the respective setup.cfg changes | 18:51 |
| adam_g | otherwise we need to bump setup.cfg to 2015.2 | 18:51 |
| *** puranamr has joined #akanda | 18:59 | |
| *** fzylogic has quit IRC | 19:02 | |
| *** fzylogic has joined #akanda | 19:04 | |
| *** puranamr has quit IRC | 19:04 | |
| *** smcginnis has joined #akanda | 19:12 | |
| *** openstackgerrit has quit IRC | 19:46 | |
| *** openstackgerrit has joined #akanda | 19:46 | |
| sarob | adam_g: im updating my semver patches right now | 20:08 |
| sarob | adam_g: my mistake for not following through | 20:08 |
| *** drwahl has quit IRC | 20:10 | |
| openstackgerrit | Sean Roberts proposed stackforge/akanda-rug: release versioning servers change https://review.openstack.org/197787 | 20:11 |
| sarob | adam_g: ^^ | 20:12 |
| *** openstackgerrit has quit IRC | 20:16 | |
| *** openstackgerrit has joined #akanda | 20:16 | |
| *** puranamr has joined #akanda | 20:21 | |
| *** puranamr has quit IRC | 20:26 | |
| *** thingee has joined #akanda | 20:31 | |
| thingee | whoa dreamhost crew in da house | 20:31 |
| markmcclain | adam_g: thingee was asking me about the ring stuff you've been working on | 20:32 |
| *** puranamr has joined #akanda | 20:32 | |
| adam_g | sorry, cant chat now will be available soon | 20:32 |
| *** puranamr has quit IRC | 20:36 | |
| *** puranamr has joined #akanda | 20:45 | |
| sarob | adam_g: no prob | 20:45 |
| adam_g | sarob, you're going to need to push 7.0.0.0b1 before the setup.cfg change will work | 20:46 |
| sarob | really, why? | 20:46 |
| *** openstackgerrit has quit IRC | 20:46 | |
| *** openstackgerrit has joined #akanda | 20:46 | |
| sarob | i tagged at 2015.2.0b1 | 20:46 |
| sarob | wouldnt the pbr version need to be 2105.2? | 20:47 |
| *** openstack has joined #akanda | 21:09 | |
| -cameron.freenode.net- [freenode-info] channel trolls and no channel staff around to help? please check with freenode support: http://freenode.net/faq.shtml#gettinghelp | 21:09 | |
| sarob | puranamr: you need to be on the gerrit change | 21:10 |
| sarob | puranamr: git fetch https://review.openstack.org/stackforge/akanda refs/changes/59/207559/1 && git checkout FETCH_HEAD | 21:10 |
| sarob | puranamr: you can find the link on the review page | 21:10 |
| puranamr | sarob: i see it | 21:11 |
| sarob | puranamr: on the patch set | 21:11 |
| sarob | puranamr: the little tiny copy to buffer icon | 21:11 |
| sarob | puranamr: not very easy to figure out | 21:11 |
| sarob | or copy the string | 21:11 |
| puranamr | sarob: i see it, i will cancel the previous review and merge it back | 21:12 |
| sarob | puranamr: cool | 21:12 |
| sarob | adam_g: you good with the 2015.2.0b1 for lm1 and then voting next week on proceeding with 2015.2.0b2 or 7.0.0b2? | 21:14 |
| sarob | its all minuta right now for our eyes | 21:14 |
| sarob | minutia | 21:18 |
| adam_g | sarob, oh, ok. | 21:20 |
| adam_g | 2015.x is fine by me | 21:20 |
| sarob | adam_g: of course gate is broken when i break the tests | 21:22 |
| sarob | adam_g: and push the fix | 21:22 |
| sarob | sigh | 21:22 |
| adam_g | sarob, huh? | 21:46 |
| adam_g | 197787 should pass | 21:46 |
| adam_g | you'll need similar patches to the other projects you tagged | 21:47 |
| sarob | adam_g: yup, | 21:47 |
| adam_g | thingee, hey wassup | 21:47 |
| sarob | adam_g: i was waiting to make sure we were in synch | 21:47 |
| thingee | adam_g: so I'm getting some pressure from folks on getting Cinder active/active. The problem is if two cinder-manager processes work with the same resource like a volume, they could clobber each other. People are suggesting tooz and other complexity that I'm not sold on Cinder needing. I had a crazy idea though. What if I just had each cinder api and | 21:48 |
| thingee | cinder-manager process subscribe to a topic. Everytime a cinder-manager does anything with a volume for example, it would broadcast the uuid. When it's done, it'll broadcast it's done with that uuid. Each cinder-manager process and cinder api process will take that broadcast and store the uuids being used in memory. The cinder api will always stop something | 21:48 |
| thingee | from happening if a uuid was broadcasted. The cinder manager will also not attempt something if it already knows that uuid is being used. None of the processes need to know about each other since they're only subscribing to a mq topic. I have not figured out what time out will look like for this yet. | 21:48 |
| thingee | adam_g: then dhellmann told me you're bring in the swift ring solution instead of me doing something wrong and crazy | 21:48 |
| sarob | adam_g: yeah, https://review.openstack.org/#/c/197787/ | 21:48 |
| thingee | adam_g: ok, I'll bb, my plane is about to land. | 21:48 |
| openstackgerrit | Rama Puranam proposed stackforge/akanda: Implements: blueprint vm-reduce-footprint https://review.openstack.org/207559 | 21:49 |
| thingee | adam_g: but please point me to anything on the efforts with akanda bringing in the swift ring solution. I'm curious | 21:49 |
| adam_g | thingee, actually its not the swift ring | 21:49 |
| sarob | adam_g: should pass, but its stuck as check has 570 jobs | 21:49 |
| adam_g | thingee, ive copied the hash ring from ironic | 21:49 |
| thingee | is this merged, in review with akanda? | 21:50 |
| adam_g | thingee, no im still working on it, https://review.openstack.org/#/c/195366/ | 21:50 |
| adam_g | thingee, but our use case is different than yours. it hink what you've describe more closely resembles how ironic uses its hash ring | 21:51 |
| thingee | adam_g: excellent thank you. I'll try to find information more on this | 21:51 |
| adam_g | (i think) | 21:51 |
| adam_g | still trying to grok brain dump above :) | 21:51 |
| thingee | ooo a gist in the commit message. I appreciate short answers | 21:52 |
| adam_g | thingee, im using tooz for tracking cluster membership | 21:52 |
| adam_g | thingee, that membership list is whats used to construct the consistent hash ring | 21:52 |
| adam_g | tooz broadcasts membership changes to all nodes | 21:52 |
| adam_g | every time it changes, nodes rebalance their ring given the new list of members | 21:53 |
| adam_g | thingee, https://github.com/stackforge/akanda/blob/master/specs/liberty/rug_ha.rst | 21:53 |
| thingee | adam_g: thank you | 21:53 |
| adam_g | thingee, im not sure my solution would offer the kind of guarantees you'd want for clustering a storage system, though | 21:55 |
| thingee | it's not about clustering storage systems, it's clustering the cinder-managers which are essential just message passers (i.e. cinder api->cinder-manager->storage solution) | 21:57 |
| thingee | and making sure to cinder managers don't do an attach while another is doing a snapshot or something | 21:57 |
| adam_g | thingee, is cinder-manager a new thing? or is it the same as cinder-volume? | 21:59 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Bump version in setup.cfg to 2015.2 https://review.openstack.org/197787 | 22:50 |
| *** cleverdevil has quit IRC | 23:00 | |
| *** puranamr has quit IRC | 23:50 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!