| SimplySeth | how does one query the api using curl ? | 00:59 |
|---|---|---|
| SimplySeth | I tried curl -u "username:password" "http://machine/MAAS/api/1.0/nodes/?op=list and got" "Unrecognised signature" | 01:02 |
| malikeye | anybody deploy openstack with landscape / maas? I can get the OS deployed on the first node, but juju just sits there waiting for something | 01:33 |
| malikeye | I have two network on the juju / landscape VM, and all of the hosts have 2 NIC's in them | 01:34 |
| malikeye | do I need to define the networks in maas? the instructions aren't really clear | 01:34 |
| malikeye | ahh, it wants to lookup the host with a hostname... gotta flip the DNS on I think | 01:36 |
| malikeye | now it's stuck on deploying landscape... one wall at a time :) | 01:58 |
| malikeye | so I have two nics on the landscape/juju VM, and I have both networks defined in maas... do I need to put that second nic on the network? do I need to put the maas on that second network too? or does it just manage the nics on the target nodes with that second defined network? | 02:06 |
| malikeye | seems to be doing better with the second nic on the nodes attached at the maas level | 02:30 |
| malikeye | meh.... gotta have 5 machines.... only got 4 :( | 02:39 |
| === zerick_ is now known as zerick | ||
| === zz_CyberJacob is now known as CyberJacob | ||
| === CyberJacob is now known as zz_CyberJacob | ||
| === racedo_ is now known as racedo | ||
| === woop is now known as Guest39074 | ||
| === Guest39074 is now known as woopwoop | ||
| === woopwoop is now known as woopw00p | ||
| === wojdev_ is now known as wojdev | ||
| johng | hi. looking for help with MAAS 1.7. Seems to be some internal communications issue after initial image download | 17:01 |
| johng | socket.error: [Errno 104] Connection reset by peer | 17:02 |
| kiko | johng, can you give me more detail? | 17:05 |
| johng | in a proxy environment, all in one on a KVM vm. All setup works, set proxy via UI, can see image downloads happen. after image downloads, i get tjhis error in the logs and cluster is out of sync | 17:06 |
| johng | see images are there on the systems.. download works fine | 17:07 |
| johng | 2015-02-04 12:07:25-0500 [-] Logged OOPS id OOPS-d8256ede274f6493daf4df634b69b74c: error: [Errno 104] Connection reset by peer | 17:07 |
| kiko | johng, can you show me a screenshot of your /images page? | 17:08 |
| kiko | what do you mean when you say "download works fine"? | 17:08 |
| johng | I can see heavy activity tomy proxy with iftop | 17:10 |
| johng | and images screen gets to 2/2 | 17:10 |
| kiko | I see | 17:10 |
| kiko | johng, can you show me a screenshot of your /images page? | 17:10 |
| johng | not sure how to get it to you via this freenode web.. | 17:11 |
| johng | it shows the 1404 amd64 image is there | 17:12 |
| johng | no custom images | 17:12 |
| johng | 1404 and amd64 are all that is selected | 17:12 |
| SimplySeth | what is the proper way to query "http://myIP/MAAS/api/1.0/nodes/?op=list" ? a regular curl -X GET doesn't work. | 17:12 |
| kiko | SimplySeth, when you say "doesn't work" what do you mean? | 17:13 |
| johng | there is a python stack with the error as well... | 17:13 |
| SimplySeth | Kiko ... the response is :Unrecognised signature: GET | 17:14 |
| SimplySeth | kiko: works in chrome just fine though | 17:18 |
| kiko | SimplySeth, are you on 1.7 or 1.5? | 17:18 |
| saltlake | experts, am using ubunt openstack-installer to install openstack, I keep getting this complaint that there are not 5 nodes that have more than one disk. I have given each node 2 disks each 20GB partions. Please advise | 17:18 |
| kiko | I think the issue is that nodegroups is the new black | 17:18 |
| kiko | no, that's not it, you're right | 17:19 |
| kiko | SimplySeth, I will tell you in a sec | 17:20 |
| SimplySeth | kiko: 1.7 ... I think is what comes with 14.04LTS | 17:20 |
| kiko | SimplySeth, no, 14.04 currently has 1.5.4 | 17:22 |
| SimplySeth | kiko: then it's 1.5 | 17:23 |
| kiko | ok | 17:23 |
| saltlake | really been stuck at this for more than a day and I don't know how to even login to the nodes since I don't have a username and password on the node that is commissioned by maas. | 17:23 |
| saltlake | Once maas commissions a node how can anyone login to it to for debug purposes.. (I need to see if the VM has 2 disks or not) pleae help | 17:24 |
| roadmr | saltlake: did you set up an ssh key in maas? if so, you can simply ssh to nodes, that key would give you access | 17:24 |
| kiko | SimplySeth, I think the issue may be the lack of a cookie. /nodes/ does not allow anonymous requests | 17:25 |
| saltlake | raodmr: I will try that right away. But if you can it is possible to check this VM xml to see if I have the 2 disks setup right ? http://pastebin.com/RbeMVGTH | 17:25 |
| SimplySeth | kiko: okay | 17:26 |
| saltlake | grep sdb5. I don't why openstack would not see this this | 17:26 |
| kiko | SimplySeth, I can reproduce by using a private browser window and going to the URL | 17:26 |
| kiko | Unrecognised signature: GET list | 17:26 |
| saltlake | and why openstack-install repeatedly complains of the node not having more than one disk | 17:27 |
| kiko | SimplySeth, I think an ansible controller for maas would be killer! | 17:27 |
| roadmr | saltlake: hm, per this, this node has hda, vda and vdb disks, right? I don't know if this will confuse maas but it certainly confuses me :) | 17:28 |
| SimplySeth | kiko: that's why I'm trying to learn the secret sauce to querying the API | 17:28 |
| saltlake | roadmr: I changed vda and vbd to hdb and hdc resp and tried as well (since /dev/sdb5 is a physical parition .. not formatted) | 17:29 |
| saltlake | roadmr: I changed bustype from virtio to idea as well for those 2 disks vba and vdc but that failed commmisioning. Do you have a sample xmls for a node you created for openstack ? | 17:30 |
| saltlake | xmls-> xml | 17:30 |
| roadmr | saltlake: sorry, I've only ever used bare metal with maas :/ no idea about virsh nodes | 17:30 |
| saltlake | Oh no!! | 17:31 |
| saltlake | ok raodmr: I will try ssh ing to see if that tells me anything | 17:31 |
| roadmr | saltlake: did you find out where to add the key? | 17:31 |
| saltlake | I had added the key in the maas preference page | 17:31 |
| saltlake | but I rebooted the server and it won't come back up.. | 17:31 |
| roadmr | saltlake: go to your account details, you'll find an "SSH Keys" section there with a button, then put your public key there | 17:31 |
| saltlake | then ssh to the IP address for that node | 17:32 |
| roadmr | saltlake: yes, once the node is up you can ssh directly to its ip | 17:32 |
| roadmr | saltlake: (or even to its blahblah.maas hostname if you're in the maas node itself) | 17:32 |
| kiko | SimplySeth, let me research and I'll post an answer to your askubuntu question | 17:33 |
| kiko | just a sec | 17:33 |
| SimplySeth | kiko: I would greatly appreciate it. | 17:33 |
| kiko | SimplySeth, meanwhile, see SimplySeth, not answering your ansew | 17:34 |
| kiko | err | 17:34 |
| kiko | SimplySeth, meanwhile, see http://blog.allenap.me/2013/06/workaround-for-uploading-files-to-maas.html | 17:34 |
| SimplySeth | kiko: cool thanks | 17:35 |
| kiko | SimplySeth, note the use of the python client api | 17:35 |
| SimplySeth | kiko: I might have go directly to python instead of playing with BASH | 17:36 |
| kiko | SimplySeth, the issue is that the API uses OAuth | 17:37 |
| kiko | SimplySeth, I will still get you a more complete answer | 17:38 |
| SimplySeth | kiko: thanks | 17:38 |
| johng | my python error with the connection resets: | 17:40 |
| johng | 2015-02-04 12:38:31-0500 [-] Unhandled Error | 17:40 |
| johng | File "/usr/lib/python2.7/dist-packages/provisioningserver/boot/uefi.py", line 134, in download_grubnet | 17:40 |
| kiko | johng, can you put the whole traceback in a pastebin? | 17:41 |
| johng | sure - how do I do that? | 17:41 |
| johng | googled it | 17:42 |
| johng | see if this works | 17:42 |
| johng | http://pastebin.com/Snh4nBAi | 17:42 |
| kiko | johng, what version of maas? | 17:43 |
| johng | ii maas 1.7.1+bzr334 all MAAS server all-in-one metapackag | 17:44 |
| johng | same for dhcp and dns there as well | 17:44 |
| kiko | interesting | 17:45 |
| kiko | johng, it seems to me that the problem is that the UEFI download is not being done through the proxy | 17:45 |
| kiko | johng, can you check to see if my hunch is correct? | 17:45 |
| kiko | yes, it seems so | 17:46 |
| kiko | shucks | 17:46 |
| johng | so my LTS image downloads, but boot does not? there is only one place to set proxy | 17:46 |
| kiko | yes | 17:46 |
| kiko | it seems to be a bug in maas | 17:47 |
| johng | is it an easy edit somewhere that I can fix it? or shoudl O gi use 1.5? | 17:47 |
| kiko | I am not sure 1.5 would work tbh | 17:48 |
| johng | very much apprecite you identifying it. much much appreciate if you can provide workaround. :-0 | 17:49 |
| kiko | johng, can you confirm first that the proxy is not getting the request for the grubnetx64.efi.signed file? | 17:52 |
| johng | I don't have access to the proxy. What can I get fro yuo from the box? | 17:54 |
| kiko | johng, what port is the proxy listening on? | 17:57 |
| johng | 8080 | 17:57 |
| magicninja | guys i need some help i got maas going fine and it runs through an install from pxe boot but once done i just come to a black screen but power is still on and i can even send control alt delete and reboot the box why cant i see anything? | 17:58 |
| kiko | magicninja, this is at the end of installation, when the node would normally power off? is this maas 1.5 or 1.7? | 17:59 |
| kiko | johng, I guess a packet capture of anything going through your default route and going to port 80 would be useful | 17:59 |
| magicninja | maas 1.7 i just installed the whole thing yesterday and it does shut down after the install but when it boots back up to my hdd it goes to a black screen | 18:00 |
| kiko | johng, because that would indicate we are trying to fetch resources directly | 18:00 |
| magicninja | thx kiko | 18:00 |
| kiko | magicninja, at which point does it go to a black screen? | 18:00 |
| johng | ok. I'm set up for that. be back in a few | 18:00 |
| johng | thanks | 18:00 |
| kiko | thanks | 18:00 |
| magicninja | after install is done it powers off then back up it boots my newly installed hdd but then just goes to a black screen | 18:01 |
| kiko | magicninja, does it for instance go through grub? do you see kernel messages? | 18:01 |
| magicninja | but the computer is still on | 18:01 |
| magicninja | i dont see grub like normal | 18:01 |
| magicninja | it doesnt show up it just goes straight to black screen | 18:01 |
| kiko | magicninja, that's weird. perhaps the grub install failed? | 18:01 |
| magicninja | but i can still press control alt delete and it will reboot | 18:01 |
| kiko | right | 18:01 |
| malikeye | landscape isn't liking the hosts in maas.... all nodes had 2 x nics and 2 scsi disks, but landscape still complains that they don't have 2 disks | 18:01 |
| kiko | malikeye, Beret might be able to help you, but I think he'd EOD now | 18:02 |
| magicninja | i tried reinstalling it from a rescue usb but no dice there either i can get to grub then but after i say boot ubuntu in grub... black screen | 18:02 |
| malikeye | lsblk certainly shows 2 disks | 18:02 |
| magicninja | reinstalling grub that is | 18:02 |
| magicninja | i have another problem too if i try to do this install from a virtual box vm it boots into the image pxeboot goes fine but about halfway through starts saying ifconfig failed giving up after 60 seconds | 18:03 |
| magicninja | any ideas there im all ears | 18:03 |
| magicninja | physical machines have first problem vms just wont go past that spot they drop me to initramfs | 18:04 |
| magicninja | just ran through the automated install process again finishing up now | 18:07 |
| magicninja | doubt there will be any change though | 18:07 |
| kiko | magicninja, the ipconfig issue is a typical networking problem related to your bridge | 18:07 |
| kiko | it's very common | 18:08 |
| magicninja | but it does let me pxeboot | 18:08 |
| magicninja | it starts the image and everything just halfway through says that | 18:09 |
| magicninja | do i have to do this on like a nat virtual network to make those work | 18:09 |
| magicninja | i dont know why it wouldnt work though | 18:09 |
| magicninja | i mean how i have it now with paravirtualized devices is how im running my server | 18:09 |
| magicninja | the maas server is a vm and i just plug a box into our network and it pxeboots right away | 18:10 |
| magicninja | with hardware it even goes through the whole process | 18:10 |
| kiko | magicninja, right, it's because dhcp is not very robust to packet droppage | 18:10 |
| magicninja | with my hardware issue could it be because the drive in the box im installing is 3TB | 18:10 |
| kiko | so if the bridge for some reason drops or delays a packet | 18:10 |
| kiko | then the ipconfig will fail | 18:10 |
| magicninja | ah | 18:11 |
| magicninja | so do it on a nat'ed virtual network and that should make the vms work? | 18:11 |
| kiko | magicninja, another situation where this happens is when the node has multiple interfaces | 18:11 |
| magicninja | oh and what is the best idrac etc to use for maas | 18:11 |
| magicninja | yeah node doesnt have multiple interfaces | 18:11 |
| kiko | magicninja, any IPMI-compliant BMC should be fine | 18:11 |
| magicninja | its a vm lol | 18:11 |
| kiko | a VM with multiple NICs | 18:12 |
| magicninja | nope just one nic | 18:13 |
| magicninja | i made sure after scouring the internet | 18:13 |
| kiko | okay | 18:17 |
| magicninja | ill let you know what my hardware box does in just a minute | 18:18 |
| magicninja | im not totally sure if this would cause a problem but... | 18:18 |
| magicninja | the hardware machine i am testing with has intel chipset | 18:18 |
| magicninja | but on the frontend when i click on the node i had it marked as amd64 after noticing this after my last attempt i changed it started it again and am almost done with the fresh install | 18:19 |
| magicninja | sure is nice to have a chat room to talk this over with someone who knows about t | 18:19 |
| magicninja | this stuff | 18:19 |
| magicninja | ill try and stick around and give what support i can | 18:19 |
| magicninja | i gotta admit with the architecture marked differently the install process sure did change | 18:21 |
| kiko | SimplySeth, http://askubuntu.com/questions/581444/how-to-query-maas-api-with-curl/581688#581688 | 18:22 |
| magicninja | after a node installs it *should* come to a login prompt right? | 18:22 |
| kiko | magicninja, yes, but you should have uploaded an SSH key at this point which will allow you to ssh in, as no user is enabled to log in with a password by default | 18:22 |
| SimplySeth | kiko: thank you | 18:22 |
| magicninja | yeah i have one out there | 18:23 |
| kiko | SimplySeth, let me know if you get stuck; I tried hard to find a good way to do it with curl but it seemed like a bit too much work | 18:23 |
| magicninja | i just dont see anything on the screen so i cant tell if i have a video problem or something is wrong wrong | 18:23 |
| kiko | magicninja, my hypothesis is that the bootloader isn't getting installed | 18:25 |
| kiko | magicninja, that would lead to the machine hanging at a blank screen | 18:25 |
| magicninja | when it ran through the process it said it was installing it | 18:25 |
| magicninja | ill pop in a live cd and reinstall it to the sda drive and reboot and let you know what happens | 18:26 |
| magicninja | and even after this latest install i still get the same thing | 18:26 |
| kiko | magicninja, yeah, boot into a usb stick and chroot into the installed partition | 18:28 |
| kiko | magicninja, a grub-install should work | 18:28 |
| SimplySeth | Kiko .. I'm perfectly comfy with Python .. thx | 18:29 |
| kiko | magicninja, having said that, AIUI for a 3TB frive it needs to be a GPT partition and UEFI boot | 18:29 |
| kiko | SimplySeth, okay -- if there are issues with the API let me know, because I don't know how many users we have that don't use the `maas` commandline client | 18:30 |
| magicninja | if its not gpt would that cause a problem | 18:30 |
| kiko | magicninja, I think 2TB+ drives require GPT to be bootable | 18:30 |
| saltlake | roadmr: I tried ssh to the client machine but it isn't possible to do so !! | 18:31 |
| saltlake | The ssh key is for the maas node not the the other client nodes that are commissioned | 18:31 |
| roadmr | saltlake: oh! maybe the key didn't get copied if deployment failed :( | 18:32 |
| saltlake | the commissioning worked fine | 18:32 |
| roadmr | saltlake: hm, all deployed nodes should have that key added to them | 18:32 |
| magicninja | apparently its not up in any way cause i cant ping or anything after it goes to black screen but keyboard still functions | 18:32 |
| roadmr | saltlake: yes, the key won't be added until they're actually started or provisioned, I think | 18:32 |
| kiko | saltlake, did you ssh as the ubuntu user? | 18:32 |
| kiko | roadmr is correct, the node needs to be deployed | 18:32 |
| saltlake | I can't "deployment" to happen since the installer can't see 2 disks on the VM client | 18:33 |
| kiko | if you have only commissioned then you won't be able to connect | 18:33 |
| roadmr | saltlake: you could "start" one node to then SSH into it and verify the disk configuration, maybe | 18:33 |
| kiko | +1 | 18:33 |
| johng | tcpdump done. lots of files donwloaded, until we get to that grubx64.efi.signed file... proxy send about 4k then sends a reset wiht TSval and TSecr in the packets... | 18:34 |
| saltlake | roadmr: How do Issh to it ? I don't have the username and password since it is pxe booted and commissioned from MAAS | 18:34 |
| roadmr | saltlake: in the maas web UI, go to the node's page, there's a "start" button there | 18:35 |
| saltlake | Ahh like that I was doing a virsh start node | 18:35 |
| roadmr | saltlake: use that to start the node, it should boot up and install the operating system, it will be left in a "ready" state I think | 18:35 |
| roadmr | saltlake: yes, once maas tells you the node is ready (or deployed, perhaps), then, from the maas server's terminal, you should be able to ssh ubuntu@NODE_IP | 18:35 |
| roadmr | saltlake: it shouldn't ask for a password then, assuming you have the keys set up correctly | 18:36 |
| saltlake | ok me trying.. it is just starting up. | 18:36 |
| saltlake | ubuntu | 18:36 |
| saltlake | sorry ignore.. wrong window | 18:36 |
| roadmr | :D | 18:38 |
| kiko | johng, interesting -- seems to me like your proxy may be causing the issue (it might for instance not like the filetype or extension) | 18:39 |
| kiko | or content | 18:39 |
| saltlake | radmr: after the node is started should it reboot ? (I am seeing the virtual window) Also should I levae network as the first boot option after the VM is commissioned ? | 18:39 |
| kiko | saltlake, the node will not reboot once started; it will be left running | 18:39 |
| kiko | saltlake, the boot option is irrelevant as we instruct the node to chain boot to the disk | 18:40 |
| magicninja | kiko are you some kinda dev or something? | 18:41 |
| saltlake | roadmr: kiko: thank you. | 18:41 |
| saltlake | roadmr: it booted up and I see this : cat /proc/partitions | 18:41 |
| saltlake | major minor #blocks name | 18:41 |
| saltlake | 253 0 20971520 vda | 18:41 |
| saltlake | 253 16 20971520 vdb | 18:41 |
| saltlake | 8 0 31457280 sda | 18:41 |
| saltlake | 8 1 31456256 sda1 | 18:41 |
| roadmr | saltlake: I'd say it has 3 disks,right? | 18:42 |
| kiko | magicninja, product owner | 18:42 |
| saltlake | roadmr: Yes it does.. so my xml is correct | 18:42 |
| saltlake | roadmr: However the ubuntu installer compains "Atleast 5 nodes need more than 1 disk) and I have 8 nodes setup exaclty like this!! | 18:43 |
| saltlake | ubuntu installer - > openstack-install | 18:43 |
| roadmr | saltlake: which installer are you using? I wonder if the vda nomenclature confuses it somehow | 18:43 |
| roadmr | saltlake: maybe it expects the disks to be sd* or hd* but I don't know the installer, where can I find it? | 18:43 |
| saltlake | roadmr: I actually put hdb, hdc in the xml but the VM shows it thus!! | 18:44 |
| saltlake | roadmr: http://pastebin.com/TM5EAzZR | 18:45 |
| roadmr | saltlake: in the xml you showed me, the target names are vda and vdb; if you want them to show as hda/hdb in the VM, you have to change the <target dev='vda' to say dev='hda' for instance> | 18:45 |
| roadmr | saltlake: ohh I see you changed it then | 18:45 |
| roadmr | saltlake: after you changed the xml, did you undefine the VMs? otherwise they'll continue to use the old configuration. | 18:45 |
| saltlake | roadmr: No I only edited and restarted | 18:46 |
| roadmr | saltlake: I think you have to virsh undefine the nodes for the new config to take effect | 18:46 |
| roadmr | saltlake: hehe I also haven't worked a lot with virsh (I suck, really), but my sequence is: virsh define blah.xml; virsh start vm-name; work work; virsh destroy vm-name; virsh undefine vm-name | 18:47 |
| roadmr | saltlake: if I just start/destroy/start the vm, it doesn't reread the xml, uses the one I had when I defined it | 18:47 |
| saltlake | roadmr: I will try that and give u the status.. .. the installer is fromthis : sudo apt-get install openstack; sudo openstack-install.. It goes through 20 min and then fails on the drive issue | 18:48 |
| saltlake | ok I will be back after undefining everything in a little bit | 18:48 |
| roadmr | saltlake: yo could undefine just one, then redefine it, and restart it with maas, just to ensure it changes as you expect | 18:49 |
| roadmr | saltlake: once you're sure, do the others | 18:50 |
| saltlake | roadmr: I just the dumped the xml for that client an I see the tag "backingstore" for those disks.. is that normal ? | 18:50 |
| roadmr | saltlake: it should be OK, just specifies which storage mechanism to use | 18:51 |
| saltlake | ok will try | 18:51 |
| kiko | johng, can you check with them? | 18:52 |
| johng | I tried a diff proxy, got same error | 18:52 |
| magicninja | where is the log written about an install of node | 18:53 |
| magicninja | var log something? | 18:53 |
| johng | I'm diggin into packet capture now to see exactly why I got a reset | 18:53 |
| kiko | magicninja, on the maas side or on the node side? | 18:54 |
| saltlake | roadmr: I just undefined the VM, redined it. restarted the node from the MAAS server. sshd to it and I still see the same thing for /proc/partitions. | 18:55 |
| roadmr | saltlake: that's weird.. | 18:56 |
| saltlake | roadmr: i did not reprovision it from MAAS.. maybe that is requierd. | 18:56 |
| roadmr | saltlake: did you destroy the vm? it should not show if you do 'virsh list' | 18:57 |
| saltlake | gosh i did not .. sorry .. | 18:57 |
| roadmr | :D | 18:57 |
| === mmcc` is now known as mmcc | ||
| magicninja | ok so i ran grub-install /dev/sda and now rebooting | 19:06 |
| kiko | magicninja, is it a GPT partition table btw? | 19:07 |
| magicninja | oh you know what i didnt think to check while i was in there | 19:07 |
| kiko | heh | 19:08 |
| magicninja | fdisk -l should show me right? | 19:08 |
| magicninja | i mean i figure that it is otherwise i dont know how it could have installed there in the first place | 19:08 |
| kiko | yep | 19:08 |
| magicninja | ok going back in to check before i boot it | 19:09 |
| kiko | it's mostly curiosity | 19:09 |
| magicninja | im curious now too | 19:09 |
| magicninja | and only takes a sec to go back and check | 19:09 |
| johng | kiko - I see some double slashes in the url being requested | 19:11 |
| kiko | johng, that shouldn't matter -- you should be able to wget it from another machine (not behind a proxy) to test. | 19:11 |
| johng | trying now | 19:12 |
| kiko | johng, double-slashes are generally folded in URLs | 19:12 |
| magicninja | yeah my /dev/sda is gpt | 19:12 |
| saltlake | roadmr: Its the same after the destroy/recommission - restart http://pastebin.com/Mfmm08cS | 19:13 |
| roadmr | saltlake: damn :) can you pastebin the complete .xml file? | 19:14 |
| saltlake | Yes It on that linkk with the /proc/paritions below the xml | 19:14 |
| saltlake | roadmr: Oh it didn't put the whole file 1 sec | 19:15 |
| roadmr | saltlake: :) | 19:15 |
| kiko | magicninja, okay, so maas got that right at least | 19:15 |
| saltlake | roadmr: http://pastebin.com/xgN4rUqf | 19:15 |
| kiko | magicninja, does it contain a UEFI partition? | 19:15 |
| johng | kiko. get network error on my local pc, as well as wget from the maas box | 19:15 |
| magicninja | no just EFI i believe | 19:16 |
| johng | http://archive.ubuntu.com/ubuntu/dists/trusty-updates/main/uefi/grub2-amd64/current/grubnetx64.efi.signed is the url | 19:16 |
| magicninja | thats prob why but how do i manipulate that | 19:16 |
| kiko | magicninja, it should work | 19:16 |
| kiko | johng, 2015-02-04 17:17:08 (376 KB/s) - ‘grubnetx64.efi.signed’ saved [1001336/1001336] | 19:17 |
| kiko | works from here behind my proxy | 19:17 |
| kiko | kiko@gasolinux:~$ wget http://archive.ubuntu.com/ubuntu/dists/trusty-updates/main/uefi/grub2-amd64/current/grubnetx64.efi.signed | 19:17 |
| kiko | --2015-02-04 17:17:05-- http://archive.ubuntu.com/ubuntu/dists/trusty-updates/main/uefi/grub2-amd64/current/grubnetx64.efi.signed | 19:17 |
| kiko | Connecting to 192.168.99.4:3128... connected. | 19:17 |
| kiko | Proxy request sent, awaiting response... 200 OK | 19:17 |
| kiko | [...] | 19:17 |
| kiko | johng, looks like a local issue. can you check with the proxy gods if they can unblock the issue? | 19:17 |
| roadmr | saltlake: something continues to be very weird, you're declaring hda, hdb, hdc in your .xml, but the partitions file shows vda, vdb, sda, which is completely different. | 19:18 |
| kiko | johng, I' m surprised you are getting a network error on your local box, though, unless you are behind the same proxy. | 19:18 |
| kiko | if you are, then you've nailed the problem :) | 19:19 |
| saltlake | roadmr: I don't get it either.. I think I should deleter the physical location of the client1 img with ha | 19:19 |
| saltlake | sda!! | 19:19 |
| saltlake | roadmr: Does having the virtio bustype matter vs ide / | 19:19 |
| saltlake | roadmr: I will actually define a completely new name and stop using client1 and use it to see if there is any diff | 19:20 |
| roadmr | saltlake: sounds good. From what I just read, you should always try to use virtio | 19:21 |
| kiko | totally | 19:21 |
| saltlake | if you don't mind sharing the source of the link.. it is not so apparent from virt-install --help | 19:21 |
| johng | thanks for helping kiko. It was failing with nework error in chrome. after you tried i tried IE and it works. now it downloads in chrome as well.. go figure. no MAAS for me. guess I'll go staight juju rather than fight the peoxy-people.. | 19:22 |
| johng | wget still fails... | 19:22 |
| johng | Proxy request sent, awaiting response... 206 Partial Content | 19:22 |
| kiko | johng, how will you go straight juju without installing the nodes? | 19:23 |
| kiko | that's not a failure | 19:23 |
| kiko | partial content means that you already have part of the file locally | 19:23 |
| johng | Read error at byte 842669/1001336 (Connection reset by peer). Retrying. | 19:23 |
| johng | get lots of those.. | 19:23 |
| johng | that's wget. | 19:24 |
| kiko | johng, and the proxy is mandatory, IOW, you can't issue requests directly? | 19:24 |
| magicninja | i just dont get it i dont know what more to check even after installing grub still no dice | 19:24 |
| johng | its in a lab. | 19:24 |
| kiko | johng, is it always at that location? | 19:24 |
| kiko | magicninja, I am a bit unclear as to what the boot process looks like on your node -- is it attempting to PXE-boot first, and then going blank? | 19:25 |
| johng | need openstack on trusty. hoping I can just juju deploy stuff without MAAS. I have cobbler there.. | 19:25 |
| johng | new to juju and MAAS if you could not tell) | 19:25 |
| magicninja | i can make it do that but as i am through the install process even if i pxe boot it just boots my hdd | 19:25 |
| kiko | magicninja, that's correct | 19:26 |
| magicninja | it connects then mass tells it to boot the hard drive | 19:26 |
| magicninja | but then just blank screen | 19:26 |
| kiko | johng, you can't juju deploy without a provider | 19:26 |
| magicninja | but keyboard still works and box still on | 19:26 |
| magicninja | just no screen | 19:26 |
| magicninja | does not respond to ping either | 19:26 |
| kiko | magicninja, hmm, to rule out anything else, does a USB install on that system works? | 19:26 |
| kiko | work | 19:26 |
| johng | kiko I saw there was a manual provider but have not researched... will that not allow juju without MAAS for openstack deploy? | 19:27 |
| magicninja | thats what im trying now but im pretty sure it will | 19:27 |
| johng | am I going to have to build openstack from scratch? (uck) | 19:27 |
| === roadmr is now known as roadmr_afk | ||
| magicninja | inside the usb install when i hit partitioning it looks like maas set up three partitions one 3TB ext4 and one swap for 2.1GB and i think i did one called biosgrub 1MB | 19:29 |
| kiko | magicninja, that seems correct, it's a UEFI install IOW | 19:29 |
| kiko | magicninja, is your BIOS set up to boot in EFI mode btw? | 19:29 |
| magicninja | hmmm not sure | 19:29 |
| kiko | johng, I think you are likely to suffer more than if you just get this file to come through :-) | 19:30 |
| magicninja | before i reinstal let me check | 19:30 |
| kiko | johng, you can set up a local mirror and tell maas to use that | 19:30 |
| kiko | magicninja, that can make a difference | 19:30 |
| johng | kiko - am I able to put the file there manually and avoid that last part of the download? | 19:30 |
| magicninja | in bios i have a pci rom priority it was set to efi compatible but the only other option is legacy rom | 19:31 |
| kiko | johng, I'm not entirely sure of the answer to that :) | 19:31 |
| kiko | magicninja, pci rom? not boot? | 19:31 |
| kiko | magicninja, https://wiki.archlinux.org/index.php/beginners%27_guide#Troubleshooting_boot_problems | 19:31 |
| johng | debmirror is on my list to learn.. is that what your referring to? | 19:31 |
| kiko | johng, well, I was actually referring to a simplestreams mirror, but I need help from a dev to give you a more complete answer | 19:32 |
| kiko | johng, well.. you could definitely hack the source and avoid the need to download that file :) | 19:32 |
| kiko | johng, look at uefi.py in provisioningserver/boot | 19:33 |
| kiko | just comment out the block from line 206 to 213 | 19:34 |
| kiko | and put the file in the right place | 19:36 |
| kiko | i.e. | 19:36 |
| kiko | var/lib/maas/boot-reources/snapshot*/grubx64.efi | 19:37 |
| kiko | that should be enough -- it's the last file of the boot resources you need to download | 19:38 |
| saltlake | roadmr: same stuff with a new client that never existed so far.. I am confused.. Do you happen to have a virt-install command line you could share on how you might set up a disk .. is it because the source directory /dev/sdc5/ /dev/sdb5 are logical volumes on a physical disk ? | 19:40 |
| johng | wow, I guess I have tried a lot. I have about 40 snapshots in there | 19:44 |
| kiko | johng, we download those once a day IIRC | 19:48 |
| johng | All mine are dated today. :-) | 19:48 |
| malikeye | you actually need 5 nodes for OS itself... the landscape / juju stuff sits on it's own node... so I'm using a huge blade to host landscape / juju | 19:51 |
| kiko | johng, gar! | 19:53 |
| saltlake | malikeye: Can you share how you created the VM nodes ? | 19:53 |
| magicninja | well after reinstalling grub i can get into grub now but when i boot ubuntu it just goes to same ol black screen | 19:53 |
| kiko | johng, I need to split, but if you still have issues, could you email maas-devel and we'll look at it overnight? | 19:53 |
| kiko | thanks | 19:53 |
| magicninja | i just know this is something stupid | 19:53 |
| malikeye | I didn't use VM's... I have UCS blades... I would've liked for landscape / juju to live in a VM though | 19:53 |
| malikeye | you can't choose where it gets deployed I don't think | 19:54 |
| kiko | magicninja, did you see the URL I posted to you? | 19:54 |
| magicninja | yeah havent gone through it yet though | 19:54 |
| johng | kiko cluster howing sync after a reboot.. | 19:54 |
| kiko | magicninja, https://wiki.archlinux.org/index.php/Kernel_mode_setting#Disabling_modesetting | 19:54 |
| kiko | johng, synced! cool :) | 19:54 |
| kiko | I need to split, catch you all tomorrow | 19:55 |
| * kiko waves | 19:55 | |
| johng | do you know if the process will go back directories and find the grub file, or do I need to keep it in the latest one somhow? | 19:55 |
| saltlake | malikeye: I have a large server and am working with VM ndose for everything. I don't care for performance. | 19:56 |
| johng | later kiko! thanks a bunch | 19:56 |
| kiko | johng, the latter | 19:56 |
| malikeye | I want the controllers to be VM and the nodes to be raw KVM | 19:56 |
| saltlake | malikeye: I really hope to have a basic openstack install done else I get fired.. but if I figure it out I will share it on a blog for sure | 19:56 |
| johng | k | 19:56 |
| malikeye | these are 32-cre 256GB blades... so a whole blade for landscape / juju is a waste | 19:57 |
| malikeye | saltlake: just make sure you have 5 nova nodes, 2 disks on each node, and 2 nics on each node | 20:07 |
| malikeye | there is no way around that, unless you do the all-in-one | 20:07 |
| === roadmr_afk is now known as roadmr | ||
| malikeye | hmm, this is weird... maas only sees half the ram on these blades | 20:10 |
| saltlake2 | Now I am quite certain that this is an issue with openstack-install and that it is unable to see 2 disks on the VM nodes and its complaint about "Atleast 5 nodes should have more than 1 disk) is unfounded. I have to be NOT THE first person who is facing this. | 20:10 |
| malikeye | you need 5 nodes dedicated to nova with 2 disks | 20:11 |
| malikeye | then you need a node for landscape / juju... you can run maas on that node I believe | 20:11 |
| malikeye | if not on that same node, another for maas | 20:11 |
| malikeye | == 7 | 20:11 |
| saltlake2 | I have 8 VM nodes with 2 disks each. But the openstack-install has that gripe | 20:11 |
| malikeye | are you using virtio? | 20:11 |
| saltlake2 | 1 of those -> landscape. But there are 7 more | 20:11 |
| roadmr | saltlake2: you never told me where to get openstack-install :) I can try to see how it checks for number of disks, see if your configuration is confusing it in any way | 20:12 |
| malikeye | I think it needs to be scsi disks.... it doesn't like virtio | 20:12 |
| roadmr | saltlake2: what confuses *me* is why you get those very weird disk names. I've never used virt-install, I have a hand-hacked .xml file :/ | 20:12 |
| saltlake2 | malikeye: I have not eplored the scsi disk thing at ALL!! | 20:14 |
| saltlake2 | that could make sense | 20:14 |
| saltlake2 | roadmr: sorry. this is how I got openstack-install http://www.ubuntu.com/download/cloud/install-ubuntu-openstack | 20:14 |
| saltlake2 | point 4 | 20:14 |
| saltlake2 | malikeye: I will try to ee if making those disks scsi helps.. I hope those sources can be logical partitions. | 20:15 |
| malikeye | it will bootstrap them and overwrite everything | 20:15 |
| malikeye | I read a bug somewhere that virtio was an issue, so hopefully that will fix it for you | 20:16 |
| roadmr | saltlake2: oh boy... it'll take me a while to look at openstack-install, but I'll get around to it | 20:16 |
| saltlake2 | roadmr: You've been most helpful.. thank you So much!! | 20:19 |
| malikeye | I can't see how any "regular" person would deploy this stuff.... so many logs to look through for strict requirements for the environment | 20:19 |
| saltlake2 | malikeye: I will explore the scsi ooption.. I hope I don't need real physical disks .. but can get away with using something virtul!! | 20:19 |
| malikeye | it should work on virtual... but the storage device that's getting reported is virtio, and it probably either doesn't look for that, or blocks that as a viable storage taret | 20:20 |
| malikeye | target | 20:20 |
| malikeye | the landscape openstack install hangs at 1% with no errors in any of the logs I can see | 20:22 |
| malikeye | hmmm, second time around it's progrssing | 20:36 |
| malikeye | get this error while installing the neutron gateway : dpkg status database is locked by another process | 20:39 |
| malikeye | not to bitch, but so much for "your own private cloud in minutes", lol | 20:40 |
| magicninja | man if this would just work it would be great but at this rate i might be spinning up a fai server | 20:58 |
| magicninja | i keep getting dropped to initramfs after one of my vms boots to maas | 21:04 |
| magicninja | says alert! /dev/disk/blahblahblah does not exist dropping to a shell | 21:05 |
| magicninja | only on my virtualbox vms though | 21:05 |
| magicninja | still cant get my hardware box to boot up and not black screen | 21:05 |
| malikeye | landscape apparently doesn't like my network setup... I have the nodes, maas, and landscape on one VLAN on nic across the environment, and then a /24 on the second nic across the environment | 21:09 |
| malikeye | this really should work | 21:22 |
| magicninja | anyone else have any ideas about getting this black screen to go away make it all the way through and just cant get passed this spot | 21:31 |
| saltlake2 | malikeye: Please explain somewhere on how u setup another virtual netwrok to make landsacpe happy.. I will face the same issue | 21:38 |
| malikeye | saltlake2: I haven't gotten it to work yet, but I have 2 VLAN's... one for infra and one for instances | 21:56 |
| malikeye | everything is on both network... VLAN1 is for landscape / maas | 21:56 |
| malikeye | which works fine... when I get to the landscape part, I select VLAN2 as the "public" network, as I really have to, it forces me to, because at least one node needs to be on that network, and the NICs are all tagged to that network anyway | 21:57 |
| saltlake2 | malikeye: I have one private network for all the client nodes is and is a 192.168.122.XX and the MAAS controller is on 192.168.122.100 so all are on the private network. I don't know how to do the second network or how it should be setup ? I could connect it to the bridge on the host | 21:57 |
| saltlake2 | Would that be acceptable | 21:57 |
| saltlake2 | ? | 21:58 |
| malikeye | not sure... I'm rebuilding at this point... putting maas and the openstack-installer on one node... hoping I can target a vsphere guest as the landscape / juju host | 21:58 |
| malikeye | trying to follow the directions as close as possible... but having to use the VLAN's, which is what is screwing things up I believe | 21:59 |
| malikeye | landscape -> maas -> UCS is working fine | 21:59 |
| saltlake2 | which instructions are you following to setup this up ? | 22:02 |
| malikeye | saltlake2: http://www.ubuntu.com/download/cloud/install-ubuntu-openstack | 22:11 |
| saltlake2 | malikeye: thanks.. I really hope I figure it out !! | 22:13 |
| saltlake2 | malikeye: I am following the same link :-D | 22:13 |
| saltlake2 | malikeye: But the darn link does not talk about the requirements of the second NIC interface.. i think if one is the provate I will put tha on the local network of the office | 22:14 |
| malikeye | exactly | 22:14 |
| saltlake2 | cool.. BTW I am new to VMs .. KVMss .. openstack but bloody determined to get this done.. the only thing I am familiar with is ceph storage!! | 22:15 |
| saltlake2 | I found a way to get iscsi on the virsh xml | 22:16 |
| malikeye | meh, WOL only works on OS's vmware tools on them, and there is no vmware integration for power management | 22:21 |
| malikeye | wow, vmware announced there own openstack distribution today | 22:26 |
| malikeye | another thing that happens, is that there is no intuitive way to "clean up".... if you need to tear down and start over again, the leases get stuck..... and the leases get sent all around, so just nuking them from /var doesn't work "clean" I don't think | 22:43 |
| malikeye | fresh maas install, and it doesn't probe them memory correctly | 22:45 |
| malikeye | looks like it fails at the landscape-client install | 23:16 |
| malikeye | I can't even tell which machine is doing what at this point... maas/landscape/juju are on one VM, which deployed another box, which deploys a bunch of containers, that try to deploy another box... it's very confusing, lol | 23:16 |
| bdx | try watching the logs....... | 23:17 |
| malikeye | which logs? there are tons of them | 23:17 |
| bdx | in .cloud-install/commands.log | 23:17 |
| malikeye | none that I can find that show any errors | 23:17 |
| malikeye | I get past that point... this is landscape running the show now | 23:18 |
| malikeye | I triaged all the way to that with logs ;) | 23:18 |
| malikeye | but once it started getting distributed, I'm lost now | 23:18 |
| malikeye | sorry, my topology isn't what I just said it was... I have maas and cloud-install on one VM, which works fine... then landscape/juju get deployed to a blade, which is where all the LXC's get started, which I can get to fine... it's when I go to deploy the actual OS cluster that it fails | 23:20 |
| bdx | I go to deploy the actual OS │ gnuoy | 23:21 |
| bdx | │ cluster | 23:21 |
| bdx | what is the actual os deployment? | 23:21 |
| bdx | os cluster? | 23:21 |
| malikeye | it's openstack... it's deploying 14.04 | 23:21 |
| malikeye | yep | 23:21 |
| malikeye | it's gotta be something with my networking | 23:26 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!