| *** esikachev has joined #openstack-sahara | 00:01 | |
| *** esikachev has quit IRC | 00:05 | |
| *** esikachev has joined #openstack-sahara | 00:10 | |
| *** esikachev has quit IRC | 00:15 | |
| *** esikachev has joined #openstack-sahara | 00:19 | |
| *** esikachev has quit IRC | 00:26 | |
| *** openstackgerrit has joined #openstack-sahara | 00:44 | |
| openstackgerrit | zhongshengping proposed openstack/puppet-sahara master: Update .gitignore https://review.openstack.org/560197 | 00:44 |
|---|---|---|
| *** dave-mccowan has joined #openstack-sahara | 00:46 | |
| openstackgerrit | zhongshengping proposed openstack/puppet-sahara master: Update .gitignore https://review.openstack.org/560197 | 00:56 |
| *** shuyingy_ has joined #openstack-sahara | 01:27 | |
| *** shuyingya has joined #openstack-sahara | 02:43 | |
| *** shuyingy_ has quit IRC | 02:43 | |
| *** dave-mccowan has quit IRC | 02:56 | |
| *** shuyingy_ has joined #openstack-sahara | 03:13 | |
| *** shuyingya has quit IRC | 03:13 | |
| *** caowei has joined #openstack-sahara | 03:33 | |
| *** esikachev has joined #openstack-sahara | 05:09 | |
| *** shuyingy_ has quit IRC | 05:13 | |
| *** links has joined #openstack-sahara | 05:29 | |
| *** esikachev has quit IRC | 06:25 | |
| *** rcernin has quit IRC | 07:06 | |
| *** tesseract has joined #openstack-sahara | 07:28 | |
| *** tosky has joined #openstack-sahara | 07:55 | |
| *** pcaruana has joined #openstack-sahara | 08:13 | |
| tellesnobrega | tosky, do you know how to create a bootable volume on UI? | 12:19 |
| tellesnobrega | I'm not sure if this is something that openstack does automagically when possible, but I'm using ubuntu-spark-160 image to create a volume and it created as not bootable | 12:21 |
| tellesnobrega | nevermind | 12:21 |
| tellesnobrega | worked | 12:21 |
| tosky | I really did not know the answer | 12:22 |
| tosky | how did it work? | 12:22 |
| tellesnobrega | I went directly to volumes, create volume and used the image as source | 12:23 |
| tellesnobrega | first I went to the image and tried creating volume from ir | 12:23 |
| tellesnobrega | it | 12:23 |
| tellesnobrega | and I booted an instance with it | 12:32 |
| tosky | I see | 12:33 |
| *** dave-mccowan has joined #openstack-sahara | 12:35 | |
| *** links has quit IRC | 13:58 | |
| *** ssmith has joined #openstack-sahara | 14:40 | |
| *** cpusmith has joined #openstack-sahara | 14:42 | |
| *** ssmith has quit IRC | 14:46 | |
| openstackgerrit | caoyuan proposed openstack/sahara master: Update auth_uri option to www_authenticate_uri https://review.openstack.org/560471 | 15:32 |
| *** knylander has joined #openstack-sahara | 16:40 | |
| *** tesseract has quit IRC | 16:59 | |
| *** mgindi has joined #openstack-sahara | 17:26 | |
| mgindi | Have some questions on how sahara engine sshs to instances for pre-configuration :) | 17:27 |
| mgindi | anyone know who the right person for that would be? | 17:28 |
| mgindi | SergeyLukjanov: Hey, I see that you're the Sahara PTL, but i've also heard that you havn't been working on it for a while. Is there someone that you can refer me to that may be able to help me? | 18:14 |
| ltosky[m] | mgindi: the current PTL is tellesnobrega | 18:18 |
| tellesnobrega | mgindi, how can I help you? | 18:18 |
| mgindi | tellesnobrega: Hey how's it going? I'm running sahara in a multinode openstack environment where we have 3 controllers and HA between them, for networking service we're using non-HA DVR. So the problem I'm facing is this: | 18:21 |
| mgindi | tellesnobrega: when Sahara-engine tries to SSH to instances for its preconfiguration (before the cluster is in active state) there seems to be 1/3 chance of it working, depending on if the network namespaces is on that controller that sahara-engine thats doing the work is on | 18:23 |
| mgindi | tellesnobrega: at least that's my understanding of the problem. Any comments/opinions/info on this? | 18:24 |
| tellesnobrega | mgindi, I'm great, how about you? so, this seems like an issue that we've seen before. | 18:24 |
| tellesnobrega | how is the set up for sahara? do the cluster instances have public ip? or is sahara fully using private ips? | 18:25 |
| tellesnobrega | from what you explained, looks like s-engine can only access instance locally to it | 18:27 |
| tellesnobrega | does this seems like a possibility? | 18:27 |
| mgindi | tellesnobrega: Im doing good thank you very much :) its using fully private ips | 18:27 |
| mgindi | yes seems so | 18:27 |
| tellesnobrega | one test that you can do is, start an instance on any of the nodes (except the one where s-engine is located) | 18:28 |
| tellesnobrega | and try to ssh from the instance where the s-engine is located using private ip | 18:28 |
| tellesnobrega | if that doesn't work, then sahara won't work with your current set up, you will either have to use public ips, or setup a namespace (https://docs.openstack.org/sahara/latest/admin/advanced-configuration-guide.html) or change your routes to allow this communication to happen within your network | 18:30 |
| *** tenobreg has joined #openstack-sahara | 18:55 | |
| *** pcaruana has quit IRC | 18:55 | |
| *** tellesnobrega has quit IRC | 18:56 | |
| *** tenobreg is now known as tellesnobrega | 18:56 | |
| mgindi | tellesnobrega: I already have done a similar test. setting 'use_namespaces = True' runs the proxy command when sshing to go through the qrouter namespace locally. I had to change the proxy_command so that it goes through qdhcp namespace, sine qrouter doesn't always exist | 18:56 |
| tellesnobrega | no luck with that? | 18:57 |
| mgindi | it did work, but still only local namespaces | 18:57 |
| mgindi | is there a way we can pass over the task from one s-engine to another that would have the namespace locally? | 18:58 |
| mgindi | i wouldn't mind trying to contributing a fix for that :) | 18:58 |
| mgindi | tellesnobrega: would that be a good idea? | 19:05 |
| tellesnobrega | mgindi, sorry, was out | 19:21 |
| tellesnobrega | mgindi, I don't think we have that | 19:23 |
| tellesnobrega | I think I understand better your scenario now, you have s-engine on each controller | 19:24 |
| tellesnobrega | is that right? | 19:24 |
| tellesnobrega | mgindi, I'm asking one of our devs to give some input on this as well, he is also an operator, so he has more experience than I do with real sahara deployment | 19:25 |
| tellesnobrega | mgindi, one fix that I can think of is to set up a proxy that all s-engine use to get to the instances | 19:26 |
| mgindi | tellesnobrega: yes you're right. I have one s-engine in each controller. | 19:31 |
| mgindi | tellesnobrega: let me know if i can get in touch with the dev | 19:32 |
| mgindi | I thought about the proxy idea, but I don't know how the proxy would have access to all the namespaces in all the controllers. Its easy to do for one network, but can't think of a way for all networks | 19:33 |
| *** jeremyfreudberg has joined #openstack-sahara | 19:40 | |
| jeremyfreudberg | i'm not sure how much of the conversation i am missing (eavesdrop.openstack.org is sometimes behind), but i wanted to say that as an operator i have only ever run one sahara-engine process, so i don't have an issue with namespace not accessible | 19:42 |
| jeremyfreudberg | but i think mgindi's use case is a very valid one which needs some attention | 19:43 |
| jeremyfreudberg | i need to talk to my coworker, i think we actually do have some way of replicated qdhcp namespaces across the two controllers maybe | 19:49 |
| mgindi | jeremyfreudberg: Hey, hope ur doing well. Most production clouds will have more than one controller with HA between them, so in my view that is an important case to handle for Sahara to be used in production. | 19:50 |
| mgindi | Let me know if there's anything I can help/contribute with | 19:50 |
| *** knylander has quit IRC | 19:50 | |
| jeremyfreudberg | yes, i think it could be solved just at the neutron level (trying to figure out how we do it in my cloud) | 19:51 |
| mgindi | jeremyfreudberg: dhcp_agents_per_network is the config setting -> https://docs.openstack.org/newton/networking-guide/config-dhcp-ha.html | 19:57 |
| jeremyfreudberg | mgindi: i have to say i'm not really a neutron expert - but in our cloud we have that config value set to 2 (because we have 2 controllers) | 19:59 |
| jeremyfreudberg | mgindi: oh, i found it, we have two other config changes we made as well | 20:00 |
| jeremyfreudberg | l3_ha should be true | 20:01 |
| jeremyfreudberg | allow_automatic_l3agent_failover should be true | 20:01 |
| jeremyfreudberg | (keep in mind, i'm just reading this from a git blame -- it might not be so relevant as i think) | 20:01 |
| mgindi | jeremyfreudberg: hmmm.. we used DVR instead of l3 HA.. | 20:02 |
| mgindi | but maybe increasing qdhcp will do the trick. Can i try that and get back to u? | 20:03 |
| jeremyfreudberg | mgindi: sure, give it a try | 20:03 |
| jeremyfreudberg | (we definitely don't use DVR in my cloud -- we have l3 agent mode = legacy) | 20:04 |
| mgindi | If this doesn't work. would we be able to figure something out for DVR? :) | 20:06 |
| jeremyfreudberg | we definitely should figure things out for DVR. i'd want sahara to work for all kinds of networking scenarios | 20:07 |
| mgindi | ok great. I'll get back to u as soon as im done testing | 20:08 |
| jeremyfreudberg | mgindi, i have to sign off, but email me at <my irc nick>@gmail.com | 20:08 |
| jeremyfreudberg | i'm not on irc as much as i should | 20:08 |
| mgindi | jeremyfreudberg: thanks a lot for your help! C ya | 20:09 |
| *** jeremyfreudberg has quit IRC | 20:09 | |
| *** mordred has quit IRC | 20:28 | |
| *** vkmc has quit IRC | 20:28 | |
| *** mordred has joined #openstack-sahara | 20:31 | |
| *** vkmc has joined #openstack-sahara | 20:31 | |
| *** cpusmith has quit IRC | 20:51 | |
| tellesnobrega | mgindi, sorry I was away again | 20:54 |
| tellesnobrega | hope jeremy was able to help | 20:54 |
| tellesnobrega | let me know how it goes | 20:54 |
| mgindi | tellesnobrega: thanks for your and jeremy's help. Will do | 21:07 |
| *** rcernin has joined #openstack-sahara | 21:49 | |
| *** mgindi has quit IRC | 21:51 | |
| -openstackstatus- NOTICE: zuul was restarted to updated to the latest code; you may need to recheck changes uploaded or approvals added between 21:30 and 21:45 | 22:30 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!