Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2014-06-06

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:08 Kupo24z1 rmoe: you familar with how fuel does its ceph snapshots using ceph ephemeral disks?
00:10 Kupo24z1 trying to script something to copy childen snapshots for backup purposes but im unsure if its possible
00:14 rmoe I'm not familiar with the internals, but nova will rely on qemu to create and manage snapshots
00:14 rmoe they'll be stored in glance
00:16 Kupo24z1 Do you know if there are any plans from moving glance snapshots from full disk images to a thin/sparse format?
00:22 Kupo24z1 and does 'Use qcow format for images' do anything with ceph?
00:22 angdraug nothing
00:22 angdraug ceph's only going to work with raw images
00:23 Kupo24z1 thats what I thought, maybe should be disabled in future builds
00:23 Kupo24z1 the checkbox, that is
00:23 angdraug you can put qcow images in glance but then you'd need to download/convert them before booting from them
00:23 angdraug with raw images and our rbd-ephemeral-clone patches, you can use copy-on-write from glance to nova
00:23 angdraug but it's not qcow2 copy-on-write, it's ceph's snapshots
00:24 Kupo24z1 I've tested that and it works well, however 'glance snapshots' still take up a raw size of the 100% empty disk
00:24 Kupo24z1 im thinking a clone of the ceph snapshot would be far better suited for that
00:28 Kupo24z1 if qcow2 images are enabled, will glance snapshots be smaller?
00:30 angdraug it still won't use ceph snapshots
00:30 angdraug there's a missing feature in nova
00:31 Kupo24z1 There a blueprint for it?
00:35 angdraug https://blueprints.launchpad.net/nova/+spec/implement-rbd-snapshots-instead-of-qemu-snapshots
00:36 Kupo24z1 hmm not much activity :( thanks for the link
00:38 Kupo24z1 I might just end up coding my own system then
00:38 angdraug yeah, nobody except Mirantis has been taking taking Nova+Ceph seriously for a while
00:39 Kupo24z1 is it possible to copy a child snapshot? Dont see anything on the ceph doc page https://ceph.com/docs/master/rbd/rbd-snapshot/
00:39 angdraug you try to use this as a starting point: https://github.com/angdraug/nova/commits/rbd-ephemeral-clone
00:40 angdraug yes, you can copy child snapshots
00:40 Kupo24z1 is that the correct way to backup/snapshot a running instance or am i in the wrong line of thinking?
00:41 angdraug copy is the wrong word, otherwise you're good :)
00:41 angdraug what you need to do is clone the rbd object that back the root drive
00:41 angdraug back->backs
00:42 angdraug and then pass URI of that clone to Glance, and make Glance create an image based on existing rbd object,
00:42 angdraug instead of downloading and re-uploading it into a new rbd object
00:43 angdraug or pass URI to Glance first, and make Glance do the cloning
00:43 angdraug first approach is better because that patch series I linked above already adds clone capability to Nova
00:43 angdraug you just need to support cloning in the opposite direction
00:44 angdraug if you end up going anywhere with this let me know, I'd love to see what you come up with
00:44 Kupo24z1 'clone the rbd object that backs the root drive', that listed in rbd ls poolname?
00:47 angdraug yes, but there's better ways to get that
00:47 angdraug it's <instance_uuid>_disk
00:48 angdraug it's also recorded somewhere in instance metadata afair
00:48 angdraug doing rbd ls is fine for diagnostics but you don't want to do that from inside nova, it's slow and buggy
00:48 angdraug e.g. it would return a non-0 exit code if the pool is empty
00:48 xarses joined #fuel
00:49 Kupo24z1 When the rbd _disk object is cloned will it be a child or its own thing, not attached to the image anymore?
00:52 angdraug rbd clones keep the clone chain until you flatten them
00:52 angdraug it's recommended not to let that chain grow more than 10 clones
00:53 Kupo24z1 wouldnt it be recommended to flatten after it hits glance?
00:59 Kupo24z1 I'm probably going to make something external to openstack right now, after the immediate need is over we can probably assign some developers to do a fork
01:02 Kupo24z1 angdraug: thanks for the help, i'll be around tomorrow with more questions im sure :)
01:09 rmoe joined #fuel
01:13 angdraug looking forward to it :)
01:13 angdraug re flattening, should be tunable, it's a tradeoff between space efficiency and performance
05:17 bookwar joined #fuel
05:19 xarses joined #fuel
05:26 xenolog joined #fuel
05:26 vkramskikh joined #fuel
05:26 holser joined #fuel
05:26 monester joined #fuel
05:26 evg joined #fuel
05:26 aglarendil joined #fuel
05:26 mrasskazov joined #fuel
05:26 kat_pimenova joined #fuel
05:26 jkirnosova joined #fuel
05:26 apalkina joined #fuel
05:26 akislitsky joined #fuel
05:26 MiroslavAnashkin joined #fuel
05:26 MorAle joined #fuel
05:26 brain461_ joined #fuel
05:26 meow-nofer joined #fuel
05:26 mihgen joined #fuel
05:26 dburmistrov joined #fuel
05:26 teran joined #fuel
05:26 nurla joined #fuel
05:26 book` joined #fuel
05:26 wayneeseguin joined #fuel
05:26 rsFF joined #fuel
05:26 gleam joined #fuel
05:26 justif joined #fuel
05:26 mattymo joined #fuel
05:26 GeertJohan joined #fuel
05:26 64MAAU9U7 joined #fuel
05:26 AndreyDanin joined #fuel
05:26 xdeller joined #fuel
05:26 zhiyan_ joined #fuel
05:30 evgeniyl joined #fuel
05:33 xenolog joined #fuel
05:33 vkramskikh joined #fuel
05:33 holser joined #fuel
05:33 monester joined #fuel
05:33 evg joined #fuel
05:33 aglarendil joined #fuel
05:33 mrasskazov joined #fuel
05:33 kat_pimenova joined #fuel
05:33 jkirnosova joined #fuel
05:33 apalkina joined #fuel
05:33 akislitsky joined #fuel
05:33 MiroslavAnashkin joined #fuel
05:33 MorAle joined #fuel
05:33 brain461_ joined #fuel
05:33 meow-nofer joined #fuel
05:33 mihgen joined #fuel
05:33 dburmistrov joined #fuel
05:33 teran joined #fuel
05:33 nurla joined #fuel
05:33 book` joined #fuel
05:33 wayneeseguin joined #fuel
05:33 zhiyan_ joined #fuel
05:33 xdeller joined #fuel
05:33 AndreyDanin joined #fuel
05:33 64MAAU9U7 joined #fuel
05:33 GeertJohan joined #fuel
05:33 mattymo joined #fuel
05:33 justif joined #fuel
05:33 rsFF joined #fuel
05:33 gleam joined #fuel
06:22 al_ex joined #fuel
07:07 Arminder- joined #fuel
07:08 artem_panchenko joined #fuel
07:13 taj joined #fuel
07:14 bookwar left #fuel
07:16 bookwar joined #fuel
08:03 e0ne joined #fuel
08:06 izinovik joined #fuel
08:16 e0ne joined #fuel
08:23 xarses joined #fuel
09:03 dilyin joined #fuel
09:58 vkozhukalov joined #fuel
10:15 e0ne joined #fuel
10:30 e0ne_ joined #fuel
10:31 vkozhukalov joined #fuel
11:03 izinovik joined #fuel
11:11 e0ne joined #fuel
11:17 e0ne joined #fuel
11:37 e0ne joined #fuel
11:52 b-zone joined #fuel
12:13 sbog joined #fuel
12:13 sbog left #fuel
12:14 sbog joined #fuel
12:20 e0ne joined #fuel
13:06 jues joined #fuel
13:14 e0ne joined #fuel
13:42 e0ne joined #fuel
13:42 vkozhukalov left #fuel
13:48 jobewan joined #fuel
14:38 e0ne_ joined #fuel
14:44 sbog joined #fuel
15:00 e0ne joined #fuel
15:04 albionandrew joined #fuel
15:05 albionandrew MiroslavAnashkin: The puppet command we talked about ran. I rebooted put I still can’t pxe boot.
15:30 e0ne joined #fuel
15:42 blahRus joined #fuel
15:53 xarses joined #fuel
16:00 sbog joined #fuel
16:01 sbog joined #fuel
16:10 decede Hello, so Fuel 5.0 isnt happy with adaptec cards?
16:12 MiroslavAnashkin decede: Master node -no. Other nodes depend on VendorID, flashed to these Adaptecs.
16:21 MiroslavAnashkin albionandrew: Please run `cobbler sync` to force Cobbler to refresh configurations.
16:23 albionandrew MiroslavAnashkin: “TASK COMPLETE” try and PXE boot now?
16:23 MiroslavAnashkin yep
16:24 crshman joined #fuel
16:25 crshman left #fuel
16:31 albionandrew MiroslavAnashkin: Using tcpdump I see the request from the client to the server but still no answer from the server.
16:42 MiroslavAnashkin Ok, then it's time to check the docker network settings. Please run `fuelmenu`, select Network Setup, select docker0, make screenshot, select the physical NIC you use for Admin network, make screenshot
16:43 rmoe joined #fuel
16:43 saibarspeis joined #fuel
16:44 MiroslavAnashkin Then go to PXE Setup tab, select the same way docker0 and your Admin PXE NIC and make screenshots.
16:44 MiroslavAnashkin And quit setup without saving!
17:04 xarses joined #fuel
17:04 mutex albionandrew: hey man, that is a new bug
17:04 mutex we encountered it yesterday
17:05 mutex and I think mirantis filed a bug because of it
17:05 albionandrew mutex MiroslavAnashkin I feel kind of special that I was one of the first  :)
17:05 aleksandr_null joined #fuel
17:05 mutex I was able to demonstrate that the DHCP packet never makes it to the cobbler docker instance
17:06 mutex lemme see if I can get you the bug id
17:06 albionandrew mutex yep thats what I see too. I reported it https://bugs.launchpad.net/fuel/+bug/1324152?comments=all
17:07 mutex here is the bug we found yesterday
17:07 mutex https://bugs.launchpad.net/fuel/+bug/1327009
17:08 MiroslavAnashkin mutex: It is very probable with the latest Fuelmenu. It became too powerful and too less documented, so I have to create user guide for just this tool only.
17:09 mutex heh
17:10 albionandrew MiroslavAnashkin: I know you and xarses etc are busy and I do honestly appreciate your help do you also have anything new on http://docs.mirantis.com/fuel-dev/develop/nailgun/development/env.html etc.
17:11 mutex albionandrew: apparently the fix is to change the interface in astute.yaml and regenerate the docker container, then restart
17:11 mutex the container
17:11 xarses albionandrew: I don't understand the context
17:11 xarses what's wrong with the develop env docs
17:12 albionandrew I tried to build the iso but had issues at http://pastebin.com/TFd1cftg
17:12 albionandrew mutex thanks I’ll try that.
17:12 xarses this is in regards to the issue with the dhcp interface?
17:13 MiroslavAnashkin xarses: A lot of people attempts to build ISO with Ubuntu 14.04
17:13 albionandrew Just so that I can rebuild the ISO with the patches.
17:13 MiroslavAnashkin Or more exactly under 14.04
17:16 casanch1 joined #fuel
17:18 xarses joined #fuel
17:19 casanch1 joined #fuel
17:19 xarses albionandrew: shotgun is in <wherever you cloned fuel-main>/build/repos/nailgun if the folders don't exist run make repos from the root of the fuel-main clone
17:20 albionandrew xarses thanks . I’ll take a look
17:20 albionandrew I think I’d like to set something up to build ISOS because we want to keep using fuel.
17:35 xarses We have an open bug of high priority to make the setup for building ISO's easier
17:36 xarses It should be easier soon
17:36 xarses back to your dhcp issue, as mutex pointed out, if you fix /etc/astute.yaml and use docker to erase the cobbler, nailgun, and postgress containers and then run bootstrap_admin_node.sh again, it should clean its self up
17:37 xarses albionandrew: ^^
17:39 albionandrew xarses mutex MiroslavAnashkin thanks for your help. In the middle of something but will try in a little bit
17:39 * mutex high five
18:00 e0ne joined #fuel
18:15 albionandrew mutex http://pastebin.com/zKg4ses7 - Are you saying change this to eth0 because the dhcrelay will deal with the rest? xarses
18:22 xarses albionandrew: also, fix the ip addresses to match eth0's range
18:22 albionandrew xarses great thanks
18:22 xarses and then if you delete the containers, and re-run bootstrap_admin_node.sh (puppet) it will clean everything up
18:23 xarses you need to delete the cobbler, nailgun, and postgresql containers
18:24 xarses 'dockerctl destroy nailgun cobbler postges'
18:35 angdraug joined #fuel
19:13 BillTheKat joined #fuel
19:18 albionandrew xarses the docker process is running… so whilst I wait I’ve gone back to http://docs.mirantis.com/fuel-dev/develop/nailgun/development/env.html step 7 - http://pastebin.com/6efEhgDi
19:19 albionandrew I assume the issue is  No module named docker etc etc how do I fix that?
21:54 bookwar left #fuel
23:08 ykotko joined #fuel
23:45 e0ne joined #fuel

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary