g3 | herp a derp | 00:03 |
---|---|---|
shahaan | Hello, since upgrading to 2.1.3+bzr5573-0ubuntu1~16.04.1 500, GPT partitioning is causing MBR corruption leading to grub-rescue prompt upon booting, any ideas ?? | 06:09 |
shahaan | Upon Booting the nodes and not mass-server... | 06:10 |
BlackDex | doesn't gpt remove the mbr with maas? | 07:15 |
BlackDex | or at least there shouldn't be mbr on the disk and only gpt? | 07:16 |
errr | is it possible to create a storage layout with multiple volume groups and multiple logical volumes in each group? | 07:35 |
miono | We're trying to install machines on different VLANs from the same MAAS-instance. But the only way we can get it to work is if we set the maas-url in the rackd-config and regiond-config to MaaS IP on that vlan. Isn't it possible for MaaS to install machines on different VLANs? | 07:40 |
=== frankban|afk is now known as frankban | ||
pmatulis | miono, yes, it should be possible. is DHCP enabled on each? what errors do you get? | 09:27 |
BlackDex | oke, i have a strange problem, i have a custom maas image which i deploy, and sometimes it gets an ip after the last reboot, and sometimes it doesn't. I can put the network config to dhcp or static what i want, but it just doesn't. I don't have curtin or cloud-init installed in the image it self, but that worked before. it seems it broke somewhere around xmas. | 09:36 |
miono | pmatulis: Yeah, DHCP is enabled on both VLANs. The machines are booting over PXE and it seems to work, but during the enlistment-phase (iirc) they are trying to fetch stuff from the standard gateway. Which is not the maas_url. | 10:07 |
miono | pmatulis: Our example is like this: We have two VLANs, VLAN 100 and VLAN 200. If we set the maas-url to 10.42.100.2 we can install machines in VLAN100, but not those in VLAN200. And vice versa. | 10:09 |
pmatulis | miono, the nodes need a route to the internet for commissioning | 10:18 |
pmatulis | so check the gateway, routes, etc | 10:19 |
miono | pmatulis: They have that. But they are requesting maas on the wrong IP. They are basically sending a request to 10.42.200.1 instead of 10.42.200.2 | 10:20 |
miono | pmatulis: I will try this again now to see exactly when the error happens and what it looks like. | 10:21 |
hachi | Hi | 10:22 |
hachi | After deploying some nodes with the LACP bond i found that the network interfaces that make up the bonding also got the settings of the LACP, like this: | 10:22 |
hachi | http://paste.openstack.org/show/597325/ | 10:22 |
miono | hachi: Yeah, we're having that issue too. | 10:23 |
miono | hachi: But I haven't found a solution. Basically just decided to live with it, since they don't affect the bond itself. | 10:23 |
miono | hachi: Sorry, nvm, we're having a different issue. | 10:24 |
pmatulis | hachi, can you open an issue on it? | 10:24 |
hachi | miono and pmatulis thank you both. | 10:26 |
hachi | pmatulis do you mean filing a bug ? | 10:26 |
pmatulis | hachi, yes: https://bugs.launchpad.net/maas/+filebug | 10:26 |
hachi | OK i am going to do it now. | 10:26 |
miono | pmatulis: For some reason it's trying to make a request to a 169.254-address | 10:28 |
miono | pmatulis: (when it's trying to call home to MaaS) | 10:28 |
miono | pmatulis: And then it starts sending requests to it's standard GW | 10:28 |
miono | 10.42.200.1 | 10:28 |
pmatulis | sounds like dhcp is not working well | 10:29 |
pmatulis | check the logs for the rackd (/var/log/maas/rackd.log?) | 10:30 |
pmatulis | also, in web UI, look in the Nodes page, 'controller' tab. are all elements/services ok (green)? | 10:30 |
miono | pmatulis: Checked the UI, everything is green there. And DHCP is enabled on vlan 200. If I go to the vlan-page for vlan200 I can see the host there. 10.42.200.10 (which is in the maas dhcp-range). | 10:32 |
miono | I can ping this address from the maas-machine too.. | 10:32 |
miono | trying agian now, while tailing the rackd-log | 10:32 |
miono | it hangs for a while at "Starting Initial cloud-init job (metadata service crawler) | 10:35 |
miono | " | 10:35 |
miono | and then these requests for the wrong IP are shown on the screen of the node | 10:35 |
mup | Bug #1661203 opened: LACP bonding resulted in uncorrect settings of the primary interfaces <MAAS:New> <https://launchpad.net/bugs/1661203> | 10:36 |
miono | But since it succeeds with PXE-booting etc, I would guess that DHCP is in fact working as it should. | 10:36 |
miono | If I change the maas_url in both rackd.conf and regiond.conf to 10.42.200.2 instead of 10.42.100.2, everything works. | 10:36 |
BlackDex | miono: you need to dpkg-reconfigure maas-rack-controller and maas-region-controller and point the URL to the IP which is used for the DHCP of the enlisting | 10:41 |
BlackDex | that will trigger that IP to be used as default/backup for metadata stuff during boot | 10:41 |
miono | BlackDex: Well, that is basically what we are doing, but just not doing it through dpkg-reconfigure, instead just doing it straight in the config file and then restarting maas. | 10:42 |
BlackDex | that is good also, as long as the ip is the one of the dhcp used for pxe-boot :) | 10:43 |
miono | Well, it has to be, since the nodes I'm trying to enlist are on that vlan. | 10:44 |
pmatulis | hmm, sounds like you guys are saying you cannot implement dhcp on multiple subnets | 10:46 |
miono | pmatulis: Well, dhcp is obviously working, or they wouldn't PXE-boot. | 10:46 |
pmatulis | right ok | 10:47 |
pmatulis | miono, well, i suggest opening a bug | 10:47 |
miono | but I'm trying that dpkg-reconfigure thing now, since I suspect that thoes URLs maybe are propagated to the DB too. | 10:47 |
miono | Alright tried that, didn't work, have to bounce in to a meeting now. Thanks for your support! | 10:51 |
mup | Bug #1661214 opened: Dashboard/Nodes pages blank <MAAS:New> <https://launchpad.net/bugs/1661214> | 11:24 |
Flint_ | Hi everyone! | 13:28 |
Flint_ | Guys, where is located the apache site configuration for the maas region controller? | 13:42 |
Flint_ | does MAAS embedde its own webserver? | 13:43 |
Flint_ | does MAAS embed its own webserver? | 13:43 |
Flint_ | I can't find anything related to maas on the apache configuration or directories. Is MAAS using twisted to create a webserver that will provide the django dashboard? | 13:45 |
mup | Bug #1661254 opened: Auto-assignment of network interface tag based on LSHW/LLDP <MAAS:Triaged> <https://launchpad.net/bugs/1661254> | 14:12 |
Flint_ | ping anyone? | 14:13 |
pmatulis | 'sup? | 15:44 |
errr | is it possible to create a storage layout with multiple volume groups and multiple logical volumes in each group? | 17:09 |
kiko | errr, I believe so -- does the UX limit you there today? | 17:31 |
errr | kiko: yes it does, and the docs are not clear on how one would do that | 17:32 |
roaksoax | errr: what's the difficulty. I just tested creating multiple volume groups with logical volumes without any issues | 18:01 |
roaksoax | ? | 18:01 |
errr | roaksoax: can you tell me how you did that? | 18:05 |
errr | roaksoax: Im not sure if it matters but I am on: MAAS Version 2.1.3+bzr5573-0ubuntu1 (16.04.1) | 18:06 |
roaksoax | errr: so after I commissed a machine with 1 disk, I unmounted and removed all partitions | 18:15 |
roaksoax | errr: then 1 selected 'sda', created 2 partitions | 18:15 |
errr | Lets say I had 20 machines to do this to. Is there a way to do it once so I dont have to do it 20 times? | 18:16 |
roaksoax | errr: no, you cannot add "custom" storage layouts at the moment | 18:16 |
errr | where it could be applied as a layout like how LVM, or Flat is a layout I could have my own custom layout>' | 18:17 |
errr | ah | 18:17 |
errr | this sucks. because I actually have 3500 nodes to deploy | 18:17 |
roaksoax | errr: you can always script this. We normally do this by scripting these type of things | 18:18 |
errr | I guess Ill poke at the source and see what it would take to implement my wn custom layout | 18:18 |
errr | script it how? | 18:19 |
errr | llike once its been commissed I would execute some cli like script to apply these changes to each node? | 18:19 |
errr | ideally I would want to have 5 layouts, then be able to apply said layout to a node based on a tag. | 18:22 |
errr | and Id like it to be hands off so it could just happen as automatically as possible | 18:23 |
errr | in my case I have 5 machine types and each one needs a custom storage layout based on what that machine does | 18:23 |
=== frankban is now known as frankban|afk | ||
roaksoax | errr: yeah you could scriipt it with the CLI | 18:45 |
mup | Bug #1661203 changed: LACP bonding resulted in uncorrect settings of the primary interfaces <curtin:Incomplete> <MAAS:Invalid> <https://launchpad.net/bugs/1661203> | 20:07 |
mup | Bug #1661427 opened: [2.1.3] Adding a tag from the Web UI sometimes deletes entire list of tags for Server during Save <oil> <MAAS:New> <https://launchpad.net/bugs/1661427> | 23:26 |
mup | Bug #1661440 opened: [2.1.3] Adding a new tag from WebUI fails after trying to add same new tag to different server <oil> <MAAS:New> <https://launchpad.net/bugs/1661440> | 23:56 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!