| === axw_ is now known as axw | ||
| mup | Bug #1709095 changed: Commissioning fails when cloud init can't find a data source <MAAS:Expired> <https://launchpad.net/bugs/1709095> | 04:17 |
|---|---|---|
| mup | Bug #1709095 opened: Commissioning fails when cloud init can't find a data source <MAAS:Expired> <https://launchpad.net/bugs/1709095> | 04:26 |
| mup | Bug #1709095 changed: Commissioning fails when cloud init can't find a data source <MAAS:Expired> <https://launchpad.net/bugs/1709095> | 04:35 |
| mup | Bug #1733285 opened: Unable to create pod with MAAS 2.3-rc2 on 17.10 <s390x> <MAAS:New> <https://launchpad.net/bugs/1733285> | 08:36 |
| mup | Bug #1733285 changed: Unable to create pod with MAAS 2.3-rc2 on 17.10 <s390x> <MAAS:New> <https://launchpad.net/bugs/1733285> | 08:48 |
| mup | Bug #1733285 opened: Unable to create pod with MAAS 2.3-rc2 on 17.10 <s390x> <MAAS:New> <https://launchpad.net/bugs/1733285> | 08:51 |
| === mup_ is now known as mup | ||
| xygnal | roaksoax: is the centos6 images in the maas repo currently broken? curtain fails trying to set grub root on our device. | 15:33 |
| xygnal | have a traceback but it doenst explain a whole lot | 15:34 |
| xygnal | it need sto set root to sdm, as the RAID1 is showing as final disk not first disk (We have jbods too) | 15:37 |
| roaksoax | xygnal: not to my knowledge. we've not made any changes other than getting the latest centos6 images | 15:45 |
| roaksoax | xygnal: also, we dont support efi on centos6 if that's what you were using | 15:45 |
| roaksoax | since at the time of enablement,what was needed for efi wasn't available in centos6 iirc | 15:46 |
| xygnal | roaksoax: we are running BIOS only on all of our stuff. I was just asking about EFI in case it might helkp our boot situation. | 15:52 |
| xygnal | roaksoax: we are using th strait vanilla image you host for centos7 | 15:52 |
| xygnal | er centos6* | 15:52 |
| xygnal | typo | 15:52 |
| xygnal | and it fails with that | 15:52 |
| xygnal | its not installing on /dev/sda which is the only reason I can imagine its having a problem | 15:52 |
| [Kid] | is there a way to use hostnames with juju and MAAS to deploy certain machine numbers to certain nodes by hostname? | 15:52 |
| xygnal | it shows root and boot selection for /dev/sdm but it fails with that | 15:52 |
| [Kid] | hopefully that makes sense | 15:53 |
| xygnal | roaksoax: i am booting centos6 iso now to see what order the disks select in. could be taht /dev/sdm is not the same once in the OS? | 15:55 |
| roaksoax | xygnal: for centos maas picks the first disk the os finds and that's where it installs | 15:55 |
| xygnal | hm but no that should still be in the ubuntu install image at that point | 15:55 |
| xygnal | roaksoax: we have code in our curtin preseed file that uses the boot device | 15:56 |
| xygnal | roaksoax: you helped us to code that actually to work around it | 15:56 |
| roaksoax | [Kid]: I think juju /may/ have a way to use a constraints based on machine name. It used to have, I haven't used juju in a while | 15:56 |
| xygnal | but it seems to only be working in centos7 | 15:56 |
| [Kid] | according to the guys in #juju tags are the only way | 15:57 |
| roaksoax | xygnal: strange. Both centos6/7 images process it exactly the same | 15:57 |
| [Kid] | not a huge deal | 15:57 |
| roaksoax | xygnal: the only thing I could imagine is a different version of software inside centos may be the cause ? | 15:57 |
| roaksoax | xygnal: can you file a bug ? | 15:57 |
| xygnal | roaksoax: hold on. I just disccovered he is using centos not custom, and our curtin code is only in custom | 16:02 |
| xygnal | so need to look at this first and make sure its being used correctly | 16:02 |
| roaksoax | xygnal: ack, ah that's it then | 16:06 |
| roaksoax | xygnal: it needs to be custom | 16:07 |
| xygnal | roaksoax: just tried my custom stuff in the centos file, still fails. I have debug output for you. | 16:21 |
| xygnal | roaksoax: pasted in private chat | 16:22 |
| xygnal | roaksoax: does ubuntu curtin not support the same options as custom? | 16:22 |
| xygnal | er | 16:22 |
| xygnal | sorry | 16:22 |
| xygnal | why do i keep using ubuntu? | 16:22 |
| xygnal | brain not working | 16:22 |
| xygnal | Can curtin_userdata_centos support the same options as curtin_userdata_custom | 16:23 |
| xygnal | i never used the ubuntu image - just brain tired. i meant the centos image vs the custom image. we have been using centos and centos preseed file to try to replicat the same results | 16:24 |
| xygnal | roaksoax: export the image, imported it as custom image | 16:39 |
| xygnal | tried again | 16:39 |
| xygnal | same failure, same error | 16:39 |
| roaksoax | xygnal: images under "centos" are not sent the storage config, so curtin doesn't do the partition | 16:52 |
| roaksoax | xygnal: custom images do get it | 16:52 |
| roaksoax | xygnal: that's probably why things fail | 16:52 |
| xygnal | roaksoax: well i just did custom and it failed exactly the same way | 16:53 |
| xygnal | so thats not it | 16:53 |
| xygnal | roaksoax: do you want to see the full trace output on that like i did for the previous one? | 16:57 |
| xygnal | at this point im considering writing a manual endpoint to turn jbod on/off before/after deployment manually | 16:58 |
| xygnal | because its the only way MAAS can cope with centos6 on our hardware | 16:58 |
| roaksoax | xygnal: does centos6 work with jbods ? | 17:40 |
| roaksoax | with your jbods ? | 17:40 |
| === torontoyes is now known as correct | ||
| correct | if I don't get the paid version of MAAS, this means I can't provision Windows systems? | 17:48 |
| kiko | correct, you can but you need to roll your own images, essentially | 18:05 |
| kiko | the image build service is commercial | 18:05 |
| kiko | so perhaps s/correct/partially correct/ | 18:05 |
| rick_h | howdy folks, I've got a long running maas on a single laptop that I've upgraded today to rc-2 and I'm getting the "One rack controller is not yet connected to the region. Visit the rack controllers page for more information." | 18:20 |
| rick_h | I've restarted the machine. As it's a single laptop it's both region/rack controller and I'm not up on what the new UX is trying to tell me here. Any hints on what's been changing? | 18:20 |
| rick_h | checking logs: https://pastebin.canonical.com/203665/ | 18:26 |
| roaksoax | rick_h: the rack controller cannot connect to the region for some reason | 18:28 |
| roaksoax | rick_h: pb both regiond.log and rackd.log | 18:29 |
| rick_h | k, checking | 18:29 |
| correct | kiko: lol. | 18:30 |
| rick_h | roaksoax: oh hmm, it's trying to connect ipv6 | 18:31 |
| correct | kiko: I'm rollin'em already with cobbler. Excited to test out MAAS. | 18:31 |
| correct | kiko: so.. when say roll my own, typycally, I have cobbler use ghost to deploy the image... with MAAS does it work in a similar fashion? | 18:32 |
| rick_h | roaksoax: so I've got the 10.0.0.1 in both rackd and regiond.conf files. Logs show: Region not available: User timeout caused connection failure. (While requesting RPC info at b'http://[::ffff:10.0.0.1]/MAAS/rpc/'). | 18:33 |
| correct | as I am using a ghost image, does MAAS offer some other ways of deploying my custom image? | 18:33 |
| rick_h | roaksoax: is it pulling from something other than the maas_url in each? | 18:33 |
| roaksoax | rick_h: is maas_url pointing to 10.0.0.1/MAAS or 10.0.0.1:5240/MAAS ? | 18:36 |
| rick_h | roaksoax: in the rackd it's sans-port, in the regiond it's with port | 18:37 |
| roaksoax | rick_h: use :5240, maas opens port on 5240, not on other ports | 18:37 |
| kiko | correct, no, MAAS uses curtin -- see https://readthedocs.org/projects/curtin/ | 18:42 |
| kiko | correct, sorry, I'm OTP so very slow today | 18:42 |
| kiko | shucks | 18:42 |
| kiko | correct, http://curtin.readthedocs.io/en/latest/ is what I really meant | 18:42 |
| xygnal | darn | 18:46 |
| xygnal | TJ's suspicion dint pan out | 18:46 |
| xygnal | after extracting a failed/stalled boot | 18:46 |
| xygnal | the checksums on the initrd and kernel match :( | 18:46 |
| xygnal | why is it stalling?! | 18:46 |
| correct | kiko,.. while installing MAAS, I got an error regarding psql | 18:50 |
| correct | I'm gussing server have not started | 18:51 |
| xygnal | you are are correct | 19:04 |
| correct | It appears there is a problem with the bootstrap install? | 19:04 |
| xygnal | you are are correct | 19:05 |
| correct | xygnal: umm... is there a post installation script for the db? | 19:05 |
| xygnal | i'm not answering your questions i'm just saying your name ;) | 19:05 |
| correct | lol | 19:05 |
| xygnal | i dont know what your problem is, i am in here as i have my own problems | 19:06 |
| correct | basically..I'm installing MAAS. I am at a section that configures postgressdb for MAAS | 19:07 |
| correct | it erors out. | 19:07 |
| correct | as it can't connect to DB | 19:07 |
| roaksoax | rick_h: /win 5 | 19:07 |
| roaksoax | err | 19:07 |
| rick_h | wheee | 19:07 |
| correct | I get this https://snag.gy/osnOIl.jpg | 19:09 |
| rick_h | roaksoax: hmm, seems because it's long running I'm having fun with old e/n/i vs new interface names/etc and that's causing me to have fun. Thanks for helping me poke at the logs/etc | 19:11 |
| correct | lol.. this is not working. i'm not sure what to do here. | 19:13 |
| correct | This question may not have been addressed: https://askubuntu.com/questions/965303/unable-to-install-maas-directly-using-dvd | 19:15 |
| === correct is now known as heyya | ||
| xygnal | no fun -o | 19:16 |
| xygnal | =p | 19:16 |
| heyya | If I skip the DB setup.. is can't it be configured later? | 19:17 |
| heyya | or do I need to do a layered install? | 19:18 |
| heyya | *can the postgre db be configured after ? if so, how so? | 19:18 |
| xygnal | if i had to take a guess heyya i would say, it needs to populate the database during install | 19:19 |
| heyya | xygnal, yeah.. but this install has been mostly unattended | 19:19 |
| xygnal | possible to spin up a database in advance, even if you dont keep it? | 19:20 |
| heyya | I saw a posting mention that I need to layer the install | 19:20 |
| heyya | like install ubuntu server with ssh, then maas controller, etc | 19:20 |
| heyya | ubuntu server with postgresql | 19:21 |
| heyya | I need to look at installer script | 19:21 |
| heyya | the installer is searching for a DB that does not exist | 19:22 |
| xygnal | TJ-: wireshark can decode the TFTP files! still trying to get a client-end capture though | 19:22 |
| heyya | crap.. just don't wanna do this again | 19:22 |
| xygnal | i thoguht it spun up the db for you | 19:22 |
| heyya | xygnal: that's what I thought, did you see the posting ? ---> https://askubuntu.com/questions/965303/unable-to-install-maas-directly-using-dvd | 19:23 |
| TJ- | xygnal: you still at that!? | 19:24 |
| xygnal | TJ-: well that was friday so i only just picked it up agian today. | 19:26 |
| xygnal | TJ-: I was able to confirm the whole TFTP image goes throguh, but without the client capture i can't verify the sum | 19:26 |
| TJ- | xygnal: seems like a month ago to me | 19:26 |
| xygnal | TJ-: it's new infra so we dont have spans setup yet. tough spot. | 19:27 |
| TJ- | xygnal: right; did you try with bootstrapping via iPXE | 19:27 |
| xygnal | TJ-: no, woudln't help us since MAAS doesn support it yet | 19:27 |
| xygnal | or are you implkying ipxe client can tell me if the checksum was good? | 19:28 |
| xygnal | wouldn't it still be force to use old PXE protocol because of server? | 19:28 |
| xygnal | you wanted me to spin up a whole second pxe server to test in there and that seems a bit overkill | 19:28 |
| TJ- | xygnal: As I said last week - if you booted the hardware from an iPXE CD-ROM/USB, so it then tries to do a PXE boot, and it works all 50 times (or however many you test for), it'd indicate the problem could be in the system firmware's PXE implementation | 19:29 |
| TJ- | xygnal: but if 1 or more of those tests fail you know there's an intermittent network issue | 19:30 |
| xygnal | TJ-: only happens in networks with switches that have ACI | 19:33 |
| xygnal | TJ-: same hardware. so we suspect not firmware. | 19:33 |
| heyya | isn't that ACL ? | 19:34 |
| xygnal | nope | 19:36 |
| TJ- | xygnal: So you're sure the issue is the network? | 19:36 |
| xygnal | ACI. cisco switch technology. | 19:36 |
| xygnal | TJ-: 99% certain, but i just found a cisco ACI bug release saying that it causes UDP/TFTP traffic to fail | 19:36 |
| xygnal | so crossing my fingers we aren't at taht ACI patch level | 19:36 |
| TJ- | xygnal: so, even if you prove a packet goes astray/is corrupted, you still won't know why or where it happens | 19:37 |
| TJ- | xygnal: oh really? I bet you are! | 19:37 |
| TJ- | xygnal: what's the bug reference? | 19:37 |
| heyya | Do you put all the MAAS components on one node? | 19:37 |
| TJ- | xygnal: 'fail' or 'drop packets' - the former suggests the connection fails. | 19:38 |
| xygnal | CSCaz88813 | 19:38 |
| xygnal | uz* | 19:38 |
| xygnal | not az | 19:38 |
| xygnal | "bridge packets dropped on network" | 19:39 |
| xygnal | find it? | 19:41 |
| TJ- | xygnal: that does look like it, but if it is down to a specific byte sequence, you'd expect the boot to fail the same way every time, surely? (Because its always the same data and presumably packet headers ) - Or does the server get a different IP each time/some times ? | 19:47 |
| xygnal | xygnal: yes it does get different IPs. There are commissions and deployments. | 19:47 |
| xygnal | they fail so re-trying grabs a new IP | 19:48 |
| xygnal | the commissions at the least | 19:48 |
| TJ- | xygnal: if it is that bug you can test it easily. Get the server booted successfully one-time, then run a UDP stream to it and compare what's sent to what's received. That way you have full control on the server-side | 19:48 |
| TJ- | xygnal: in other words; no need to try debugging TFTP, just debug a continuous UDP stream | 19:49 |
| TJ- | xygnal: you can use netcat (nc) either side, e.g: "cat /some/large/text/file | nc -4u server $testport" ->>> "nc -4ul $testport | tee /tmp/received/file" | 19:51 |
| xygnal | TJ-: so a netcat test? | 20:00 |
| xygnal | damn you internet | 20:01 |
| TJ- | xygnal: I think we both lost connection! did you see my netcat example? | 20:21 |
| TJ- | 19:51:36 TJ- | xygnal: you can use netcat (nc) either side, e.g: "cat /some/large/text/file | nc -4u server $testport" ->>> "nc -4ul $testport | tee /tmp/received/file" | 20:21 |
| roaksoax | heyya: not from the cd. The bug is fixed but dind't make it into the CD | 20:22 |
| roaksoax | heyya: also, you can install directly after ytou install from cd | 20:22 |
| xygnal | TJ-: we are at 2.3 fw level so, oh well | 20:22 |
| xygnal | can't be the cisco bug | 20:22 |
| TJ- | xygnal: no harm doing the UDP streaming test though. | 20:23 |
| xygnal | oh yes I have to in order to prove the network that its not my problem | 20:23 |
| xygnal | thats next | 20:23 |
| mup | Bug #1733411 opened: MAAS should create static leases for ephemeral boots when possible <MAAS:Triaged> <https://launchpad.net/bugs/1733411> | 20:49 |
| mup | Bug #1733420 opened: virsh power type check error <MAAS:Incomplete> <https://launchpad.net/bugs/1733420> | 21:19 |
| heyya | roaksoax: I just did a fresh install of ubuntu server with dns,postgres and ssh. Server is up | 21:38 |
| heyya | roaksoax: apt-cache has the package, but I was reading I should add ppa:mass/stable | 21:38 |
| heyya | anyway.. will install use the ppa | 21:41 |
| roaksoax | heyya: the ppa wont gfive you anything different from the archives | 22:14 |
| roaksoax | heyya: unless you want something newer | 22:14 |
| roaksoax | heyya: maas 2.3 goes out officially tomorrow and 2.2 will be replaced in PPA | 22:14 |
| roaksoax | so if you want the latest ppa is probably the best | 22:14 |
| heyya | sweet.. did it.. FYI MAAS is SOOOO nice | 22:15 |
| roaksoax | glad to hear that :) | 22:15 |
| heyya | very mature from what I can see | 22:15 |
| heyya | I have dhcp funning already.. I guess I can configure my dhcp to use the pxe from MAAS | 22:16 |
| roaksoax | yu can, yes | 22:18 |
| heyya | not sure how I want to set it up yet. With MAAS, I'll have 2 pxe servers running. | 22:33 |
| heyya | I guess I'll have to disable the other pxe while I'm testing | 22:33 |
| heyya | or restict it. | 22:34 |
| heyya | hmm.. ok.. I'll lock it down to a network | 22:34 |
| heyya | What are DHCP snippets in Maas? does that indicate the network for serving dhcp? | 23:11 |
| heyya | When i dig @MAAS i get WARNING: recursion requested but not available | 23:27 |
| mup | Bug #1733442 opened: Cannot delete Virsh Pod <MAAS:In Progress by newell-jensen> <https://launchpad.net/bugs/1733442> | 23:53 |
| mup | Bug #1733442 changed: Cannot delete Virsh Pod <MAAS:In Progress by newell-jensen> <https://launchpad.net/bugs/1733442> | 23:56 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
