Monday, 2011-11-21

KiallThat should have the fix I was suspecting included.. can you pastebin this file from the compute node? /usr/share/pyshared/nova/virt/libvirt/connection.py00:00
Kiall(the "pastebinit" command/package is handy for that kinda stuff .. ;))00:01
zykes-what's the discussion now on RAX stuf ?00:01
*** rsampaio has quit IRC00:02
*** agoddard has joined #openstack00:04
*** rsampaio has joined #openstack00:05
*** po has quit IRC00:06
KiallYea, That has the fix I'm thinking of alright00:07
KiallAnd you say nova-compute.log has no errors?00:09
stevegjacobsnot on controller - checking compute node now00:11
JesperA <--- What is the point of that many partitions per drive?00:13
stevegjacobsonly thing in either nova-compute.log after running the snapshot command from dashboard is this every two minutes:2011-11-21 00:11:54,868 INFO nova.compute.manager [-] Updating host status00:13
Kiallyou should see this straight away: "2011-11-21 00:16:11,323 AUDIT nova.compute.manager [bla bla admin 1] instance 1101: snapshotting"00:17
KiallI don't think there is anything that should prevent that from showing, since the "hard part" of the snapshot happens after that.00:17
stevegjacobsThis is the command I am running to initiate snapshot from the command line00:18
stevegjacobs nova image-create 5 KORE00:18
stevegjacobsThe server has either erred or is incapable of performing the requested operation. (HTTP 500)00:18
Kiallcan you try with --debug and pastebin the output?00:19
viddKiall, so you say that the issue im having has to be with keystone talking with nova and glance00:22
Kiallvidd: it sounds that way anyway...00:25
Kiallstevegjacobs: it looks like its nova-api that is failing.. strange.. anything in nova-api.log on the controller?00:28
Kiallwhat kind of image are you snapshotting? is it the standard ubuntu image?00:29
*** hugokuo has joined #openstack00:31
*** rustam has quit IRC00:33
*** rustam has joined #openstack00:34
*** dubenstein has joined #openstack00:34
dubensteinhi #openstack00:34
dubensteinis it possible to integrate keystone with nova 2012.1-dev00:35
dubensteinthe syntax of api-paste.ini changed a bit00:36
vidddubenstein, it should already be integrated with next year's essex00:36
dubensteinMetadataRequestHandler refactored00:36
*** RicardoSSP has quit IRC00:36
dubensteinso the old api-paste.ini doesn't work00:36
dubensteinvidd: so it's not an option for now ?00:37
vidddubenstein, you asked about next years release00:37
dubensteinnova-manage version says 2012.1-dev00:38
vidd2012=next year00:38
dubensteinvidd: suggest me a version to flawlessly integrate keystone, glance, nova00:38
*** rsampaio has quit IRC00:38
viddyou are working with the experimental essex release00:38
vidd"flawless" hehe00:39
dubensteinvidd: «flawless» ;)00:39
* vidd is trying to aquire that "flawless" exprience00:40
Kiallvidd: still having issues with KS+Dash?00:40
viddKiall, yes00:40
viddi can find no issues anywhere00:41
KiallOkay - Lets take dashboard out of the picture00:41
vidd(except that horizon still cant see nova)00:41
Kiallremember my script?00:41
Kiallafter RAW_TOKEN=bla00:41
Kialladd 2x lines:00:41
Kiallecho $RAW_TOKEN00:41
Kiallthen.. ./ | json_xs00:42
viddobtain token gives me the auth token00:42
Kiallsure, but the RAW_TOKEN field has lots of lovely JSON in it, including the serviceCatalog that dashboard makes use of00:43
*** rsampaio has joined #openstack00:44
viddThe program 'json_xs' is currently not installed.00:45
Kiallheh ..00:46
viddis this supposed to be installed already by something?00:46
Kiallwhatever, it just pretty prints the JSON00:46
stevegjacobsKiall: Here is a paste from nova-api.log after trying to create a snapashot of a small instance:
Kiallas good..00:46
Kiallstevegjacobs: let me compare that to my logs, maybe there is a difference in one of the entries leading up to the error00:47
*** krow has joined #openstack00:49
Kiallstevegjacobs: okay, so it looks like glance is the issue...00:49
vidd ./ | json_xs | pastebinit FTW00:50
stevegjacobsnot surprised -  the instances I've been using are the ones from before the time I used your scripts to upgrade the system00:50
Kiallstevegjacobs: humm00:50
Kiallmaybe that is related00:50
*** kaz_ has quit IRC00:51
Kiallstevegjacobs: can you wget ?00:51
stevegjacobsthey launch ok and I've continued to use them00:51
Kiallvidd: also.. it still says RegionOne00:51
stevegjacobsI think that is the one I've always used00:51
Kiallstevegjacobs: Yea, but maybe something is still odd...00:51 does my nova-config00:52
Kiallstevegjacobs: probably not, most people used the .tar.gz version (which is the "best" version to use, as you cant update it kernel.. ever..)00:52
Kiallwhich is NOT the best*00:52
Kiallvidd: ah okay ..00:52
dubensteinreverted to diablo3, euca-tools do work, but when I try to run an instance it stays pending, i found "ImageNotFound: Image 10 could not be found." in logs00:52
Kiallstevegjacobs: anyway, check the glance logs to see if there is an issue there?00:53
viddso something is messed up with my keystone_template00:53
dubensteinhelo would be appreciated :)00:53
Kiallvidd: can you pastebin that tables contents? Something is wrong in there00:54
stevegjacobsI just wget that image -  what did you want - what did you want me to do with it?00:54
vidddubenstein, i cant answer you because i have yet to get keystone to behave properly myself00:54
Kiallstevegjacobs: upload it now that you are using keystone, and try to snapshot...00:54
*** tokuzfunpi has quit IRC00:55
dubensteinvidd: novaclient works, euca-tolls works, is it keystone issue ?00:55
dubensteinvidd: i think it's something on nova-compute side00:56
pvoKiall: actually, none of the slicehost code is openstack, though a lot of the lessons we learned are in there.00:56
vidddubenstein, for your issue or my issue?00:56
Kiallstevegjacobs: save this as ""
pvoKiall: referencing previous questions....00:56
Kiallstevegjacobs: then this as whatever you want..
Kiallchmod +x the 2, and run the second in the same folder as the .img.. it will upload the image00:57
dubensteinvidd: for mine00:58
vidddubenstein, if i ever get my keystone to work, then perhaps  i can help you00:58
dubensteinvidd: i'll try, tell me more00:59
* Kiall has maintenance schedule to keep.. 00:59
viddbut you are using experimental version (aka not guarentied to work)00:59
Kiallbut vidd, those look wrong00:59
*** koolhead17 has joined #openstack00:59
dubensteinvidd: i have keystone, glance, nova working now, but heave some (i think) permission issues with libvirt01:00
koolhead17hi all01:00
Kiallspecifically, there are 2 rows with %tenant_id% and all 3 have is_global=101:00
Kiallvidd: oh and.. the URLs are completely different ;)01:01
viddok...thank Kiall i will investigate this further.....01:01
viddhave to take the boy back to his mother's so be back in a bit01:02
agoddardreading the docs, it seems like nova-volume creates an iSCSI target for the compute nodes to use, wondering how this works if you have an iSCSI SAN instead?01:03
stevegjacobsumm -I seem to have glance / keystone problems -  can't upload image even though I am sourcing my settings file.01:05
stevegjacobsglance add name="ubuntu_11_10_64bit" is_public=true < oneiric-server-cloudimg-amd64-disk1.img01:05
stevegjacobsFailed to add image. Got error:01:05
stevegjacobs401 Unauthorized01:05
stevegjacobsThis server could not verify that you are authorized to access the document you requested. Either you supplied the wrong credentials (e.g., bad password), or your browser does not understand how to supply the credentials required.01:05
stevegjacobs Authentication required01:05
stevegjacobsNote: Your image metadata may still be in the registry, but the image's status will likely be 'killed'.01:05
*** livemoon has joined #openstack01:06
*** kaz_ has joined #openstack01:08
Kiallstevegjacobs: the glance command doesnt make use of the env settings.. did you use the 2x scripts i pastebin'd? They are slightly modified from my openstack-setup scripts and handle manually getting a token from keystone etc01:09
Kiallstevegjacobs: as "" and as whatever you want..01:10
Kiall(edit the settings in the first)01:10
Kiallstevegjacobs: be back in a few mins, upgrading firmware on our cisco FW's, so my VPN is going to flake out!01:15
*** lvaughn_ has quit IRC01:16
*** Kiall has quit IRC01:17
*** Kiall has joined #openstack01:17
*** lvaughn has joined #openstack01:17
stevegjacobsKiall; nice script but there is an error somewhere
Kiallstevegjacobs: looks like keystone is giving an error rather than a token back...01:21
Kiall(There is exactly 0 error handling ;))01:21
*** JesperA has quit IRC01:22
Kiallgrab the one nova uses from the keystone api-paste and use that..01:22
Kiallit should be admin-token = something in that .ini01:22
Kialljust drop it in where the TOKEN=`./`line is01:23
Kiallbrb again, secondary FW about to kick me off again ;)01:23
*** Kiall has quit IRC01:24
*** lvaughn has quit IRC01:25
*** JesperA has joined #openstack01:25
stevegjacobsI think I  need to quit for the night - I need a snapshots of an already  running instances01:25
*** Kiall has joined #openstack01:26
*** lvaughn has joined #openstack01:30
*** rsampaio has quit IRC01:31
*** ejat has joined #openstack01:32
*** ejat has joined #openstack01:32
*** troya has joined #openstack01:34
troyahi all01:34
*** rods has quit IRC01:38
*** obino has joined #openstack01:39
smoserKiall, i'm here now if you had questions about loader kernel01:41
Kiallsmoser: yup :) still around in 5? Just finishing up a round of patches01:42
Kiallcool - thanks01:42
troyahi KiAll01:45
Kiallsmoser: back..01:46
KiallThere is an issue with OpenStack (probably UEC as well) re updating kernels that, if i remember right, you suggested the fix as to use your loader rather than the stock kernel..01:47
troyakiAll, i have question about bridge on nova, must i configure bridge on single server installation ? how if not ?01:47
KiallSo... I had 2 questions they are in a folder in "lucid-loaders", does that mean they are only for luicd? or just badly named?01:48
smoserbadly named :)01:48
smoserthanks for pointing that out.01:48
smoserthe kernel that is used is a lucid kernel.01:48
KiallAnd secondly, How does this kernel allow the instance to use a kernel from inside the image? (I'm just trying to get an understand of what i'm doing, noy how to do it! If you get me..)01:49
smoseryes, it does.01:49
smoserso there is a kernel and a ramdisk there.01:50
Kialltroya: that depends, the Flat and FlatDHCP managers need a bridge.. VLAN creates the bridge by itself.01:50
smoseryou do not want the ramdisk with '-sdb' on it, that is really there for loading a root kernel/filesystem on /dev/sdb.01:50
*** stevegjacobs has quit IRC01:51
KiallI was wondering what that one was for :)01:51
smoserbasically that thing allows you to do something like EBS root via manually starting an instance, attaching a volume with a root FS on it.01:51
troyakiAll: i use FlatDHCP on my single server01:51
smoserthe -sdb one waits 5 minutes for a /dev/sdb to show up, and then will boot off it.01:51
smoserfor th eon-sdb one.01:51
Kialltroya: then yes, nova requires you to create the bridge..01:52
troyakiAll: but i have problem when running instances, every instances which i run not running well, they only stop on BUILD state01:52
smoserkernel boots, ramdisk loads, ramdisk that is provided finds a root filesystem, reads /boot/grub/menu.lst (with awk, so not perfectly), loads the kernel and ramdisk specified there with kexec, and off you go.01:52
Kiallsmoser: ah okay, so it basically pivots the kernel out with one from the disk?01:53
*** pradeep1 has joined #openstack01:53
smoserwell, not pivot, kexec01:53
smoserso 2 kernels actually boot in this process01:53
smoserbut in order to use a kernel from the inside the image, thats kind of necessary.01:53
viddKiall, thanks for the help...that was the issue =]01:54
KiallI'm not familiar with kexec.. will have to google it in a few! So, does having 2 kernels around have any side affects or performance consequences?01:54
smoserfor ubuntu images prior to 11.04, you should use this mechanism to boot.01:54
smoserthere are not 2 kernels "around"01:55
viddhorizon now sees nova and glance -]01:55
smoserthe second completely replaces the first. the first just acks as a bootloader really.01:55
KiallOkay - I'll read up on how kexec works rather than annoy you with more questions :) The01:55
smoserthe first takes in my experience ~ 5 seconds to boot and load the next kernel, so you have that cost01:55
Kiallvidd: and does it actually work? ;)01:55
viddKiall, give me a bit to test =]01:56
*** ldlework has quit IRC01:56
viddi need to add an image now =]01:56
Kiallsmoser: thanks, I think I get it now.. I always hate using something I don't understand! Otherwise, it breaks and I have *no clue* what is going on :)01:56
smoserKiall, on eucalyptus in 10.10 i added a similar work around that uses grub and a floppy disk.01:57
viddKiall, so then you know how i feel about not "just use your scripts and move on" =]01:57
smoserhad i thought of this solution then, i would have used it.01:57
Kiallsmoser: BTW.. Semi-related, should a note about A) using a loader and B) using the .img versions for 11.04+ not be added to (You're involved directly with ubuntu right?)01:58
KiallI've noticed the vast majority of people grab the .tar.gz version and probably don't realize its not optimal...01:58
*** rustam has quit IRC01:58
smoserKiall, yeah, there should be somethign somehwer that should get you this information :)01:59
smoserand you're right about the .tar.gz01:59
smoserlargley that is just because old documentation was written prior to there being a .tar.gz01:59
smoserer... prior to .img in qcow format01:59
KiallFair enough .. Actually, 1 last question, is there ever really a need update the loader image?02:01
smoserKiall, in 12.04 there is a tool called 'cloud-publish-ubuntu' (which you could easily pull back to other releases) that just does what it should do02:01
*** mrjazzcat has joined #openstack02:01
Kiallah nice :)02:02
smoseryou still have to know to pass '--disk', but hopefully before 12.04 i can make that all "just work". the difficulty is then i have ot know osmething about your target cloud (eucalyptus wont support the .img file, neither will ec2)02:02
KiallYea.. Thats probably going to be awkward :) I guess you can sniff env var's like NOVA_URL! Bit of a hack though02:04
viddKiall, why did you switch from raw to qcow2?02:04
*** ejat has quit IRC02:04
smoserthe other thing for 12.04 time frame is that we're hoping we can get02:05
Kiallvidd: read myself and smoser's conversation ;) Its related.02:05
smosera.) in ubuntu packages, some subset of ubuntu images automatically imported into your cloud (which could be disabled)02:05
smoserb.) in ubuntu packages awareness of an ubuntu public glance server that would then just show up in your dashboard on installation.02:05
smoserie, click a button, get our images.02:05
viddKiall, im trying...but it feels like im watching a series finally and never saw any of the previous 5 seasons =]02:06
*** stevegjacobs has joined #openstack02:06
Kiallsmoser: B sounds interesting, I wonder if it would be "as simple" as adding an  additional glance endpoint to keystone..02:07
*** winston-d has joined #openstack02:07
*** ejat has joined #openstack02:07
*** ejat has joined #openstack02:07
smoserthats the general direction we're hoping for. at least at the initial point of conversation the dashboard would need to have support for multiple glance endpoints.02:08
smoserbut i really have'nt looked at this at all.02:08
viddKiall, smoser we could "create" a "ubuntu endpoint" and a02:09
viddmeathod for dasboard to pull from there and put to local galnce02:09
KiallYea, I know it can take multiple endpoints from keystone, but I dont know if it simply picks one or if it merges list the images02:09
KiallEither way, It would be interesting to see support for that, you could see all sorts 3rd party image servers popping up.. eg 10gen with a mongo DB image server etc02:10
*** chomping has quit IRC02:11
smoseri've not really looked at all at it, so i'm not qualified for a conversation at the moment. i talked to jay pipes and he is interested in helping us get there.02:11
KiallI guess the harder part of all this, as always, is political ... Managing the trust..02:11
smoserwell, you'll have the ability to verify gpg signatures02:12
smoser(per jay)02:12
smoserthat was a request/requirement02:12
smoseron ubuntu, you have our signing keys, or at least we have a safe way to get them to you.02:12
smosers/signing keys/public keys/02:12
smoser(hopefully you dont have our signing keys)02:12
viddsmoser, im sure you are aware of Kiall 's PPA's...any chance they can get reviewed for inclusion into 11.10's official repos?02:12
viddsince stock keystone and dash ont work?02:12
smoserthen you can trust the source, but figuring out what the *right* source is for something is a different story altogether02:12
smoserand one that amazon has really not made an effort at02:13
*** mrjazzcat has left #openstack02:13
smoseri dont' know about Kiall02:13
smoserkillall's ppas02:13
smoserbut if there are specific issues that can be addressed, we can/should get them addressed.02:13
smoseri think maybe i made an arse of myself earlier with similar discussion02:13
Kiallmy PPA packages are basically just updated versions of everything, not so much packaging errors.02:13
*** rsampaio has joined #openstack02:14
smoserwe *can* get things fixed in ubuntu02:14
smoserbut the fixes to stable releases are bound to SRU guidelines02:14
smoserso they need to have bugs, and cleanly defined changes.02:14
viddi think the issue is that diablo was released too close to the ubuntu freeze so patches didnt make it into the official repos02:15
viddand i fear the same thing is going to happen in 12.0402:15
Kiallsmoser: well .. the 2x biggest issues are keystone is a pre-release 2011.3 tag version, and dashboard has a few bugs in its python inclute paths02:16
smoserzul is really the person to be talking to about this stuff.02:16
smoservidd, well, the story is that essex "freeze" is supposed to be much better advertised and a much more real "freeze"02:17
smoserso that the essex release should actually work02:17
smoserthen there is time for ubuntu to get that into 12.0402:18
vidd12.04 ubuntu is a LTS so its important to get this right this time =]02:19
KiallYea - I'll be glad to see a full working set in stock ubuntu repos :) At the moment, only nova+glance+swift(I assume) work.. dashboard/keystone are broke02:19
Kiall(never used swift myself..)02:19
smoserclearly that is our goal also.02:20
Kiallsmoser: BTW you might know this.. you know that "try ubuntu" in a browser thing, is that backed by EUC/OS instances or something?02:20
smoserbut we depend on a functional essex releease :)02:20
viddsmoser, is that a "recommds" or a "requires" =]02:21
smoseri think that one is Depends.02:21
smoserKiall, runs instances on EC2.02:21
*** jog0 has joined #openstack02:22
smoser is just html mockup02:22
smoser(i'm probably terribly over-simplifying someone's work on 'take-the-tour', but the point is its not a real live thing)02:22
KiallHumm, I was sure there was a Ubuntu Desktop version of ..02:23
Kialland that take-the-tour is actually pretty impressive for HTML+CSS+JS ;)02:24
*** jkyle has joined #openstack02:26
* Kiall will be even more impressed the the thunderbird mockup actually sends me the email I just did ;)02:27
Kiallif the*02:27
*** pradeep1 has quit IRC02:28
*** troya has quit IRC02:32
*** troya has joined #openstack02:38
*** rsampaio has quit IRC02:39
viddin the "use data" field in dash, can you put ese/password info?02:46
Kiallwhatever you put there is accessible via the address02:49
KiallIf you are using ubuntu images, they come with cloud-init which has some decent stuff for parsing the data..02:49
viddKiall, i must have missed something with this vnc thing02:50
*** hugokuo has quit IRC02:51
viddi clicked the link for VNC console, and instead of getting the never-loading-page as normal, i immediately get a fail page referencing trying to access "
viddthe port is correct, but the ip is not02:51
viddwhere do i set that?02:51
*** rnorwood has joined #openstack02:51
Kiallthat should be the nova-vncproxy node02:52
Kialls/be/point to/02:52
*** jog0 has quit IRC02:52
KiallBTW, this is the kind of thing you can do with cloud-init..
Kiallif the data starts with #cloud-config , it gets handed off to a tool which does a pile of work..02:53
Kiallor if its starts with #!/bin/bash .. it gets ran.. etc etc etc02:53
*** dendrobates is now known as dendro-afk02:53
*** chomping has joined #openstack02:54
viddKiall, eveything is running on one machine so the ip is correct in the nova.conf flag02:56
Kiallwhat is it set to in nova.conf02:56
Kiallyea.. thats not a valid URL02:57
viddi missed the "/"02:58
Kiallmisplace the / ;)02:58
*** osier has joined #openstack02:58
viddim going blind looking at all these things =]02:58
viddi dont know what is scarier...these simple mistakes or that i know exactly which script to update when i fix them =]03:00
*** JesperA has quit IRC03:01
viddso...i replace with the proper IP and i get my vpn =]03:02
*** agoddard_ has joined #openstack03:05
viddthis key wants a passphrase?03:06
viddi dont understand =\03:06
Kiall"this key" ?03:07
viddi shouldnt need a passphrase for a keypair created in horizon...should I?03:07
KiallEh, it shouldn't03:08
KiallAt least, I dont remember it needing one03:08
Kiallbut i imported my existing key pretty quickly ;)03:08
viddi must have screwed up the xfer from the win-blows machine to the linux box =\03:08
*** hugokuo has joined #openstack03:08
*** freeflying has quit IRC03:12
*** freeflying has joined #openstack03:13
dubensteinmanaged to setup keystone, nova, glance and horizon. Does horizon integrate ajax-console ?03:14
viddin theory03:15
dubensteinvidd: horizon authorizes to keystone03:15
*** sdake has joined #openstack03:16
koolhead17vidd: hey there03:19
*** troya has quit IRC03:22
viddyep...i wtf'd the transfer of the pem =\03:23
Kiallvidd: BTW .. Re that convo around loaders...
KiallDoing it that way means pre 10.04 images that dont provide a .img version can have their kernels updated..03:25
viddhave no idea why id be using a pre 10.04 image...but hey...owesome =]03:25
Kiall(aka the latest LTS)03:26
*** troya has joined #openstack03:28
viddso 11.04 is already ok but the last LTS is not03:28
viddok...THAT makes sence =]03:28
Kiallif you use the 11.04 / 11.10 .img download, then yes. its fine03:28
Kiallif you use the .tar.gz then no. its not.03:28
* vidd is investigating the "custom build" option03:29
viddneed to make sure i can update a kernel image that way...or are kernels "modular"03:30
KiallIt all depends ;)03:30
Kiallif you need to add the image with "glance add ... bla bla... kernel_id=1" then you cant update the kernel03:31
viddwhen that bridge burns...ill look ya up then =]03:31
Kiallunless that kernel_id is the loader kernel ;)03:31
*** rsampaio has joined #openstack03:31
viddso a "loader" kernel will be more of a grub...saying "image 412 = 12.04/image 511 = 12.10 etc03:33
*** koolhead17 has quit IRC03:33
viddor did i miss the boat03:33
vidd[and the whole shipyard]03:33
Kiallsomething like that .. yea03:34
viddTHAT is cool03:34
*** binarychef has joined #openstack03:35
*** jkyle has quit IRC03:36
*** agoddard_ has quit IRC03:36
*** pradeep has joined #openstack03:38
Kiallanyway.. I'm off.. cya03:38
*** rsampaio has quit IRC03:41
*** troya has quit IRC03:42
*** jog0 has joined #openstack03:53
*** redconnection has quit IRC03:53
*** thickskin has joined #openstack03:58
*** rnorwood has quit IRC04:04
livemoonbye,Kiall vidd04:08
viddhey livemoon04:08
viddhow goes it04:08
livemoonI seen Kiall say "off"04:09
livemoonI think it's time for you to bed04:09
viddim getting ready to head off to bed myself04:09
* vidd has had enough frustation for a weekend =]04:10
viddits 23:10 here04:10
*** lzyeval has joined #openstack04:10
vidd[11:10 pm]04:10
viddi have a meeting in the morning04:11
*** jog0 has quit IRC04:13
*** Ryan_Lane has quit IRC04:16
*** rsampaio has joined #openstack04:22
*** dubenstein has quit IRC04:22
*** mjfork has quit IRC04:24
*** jkyle has joined #openstack04:30
*** troya has joined #openstack04:35
troyahi all04:35
*** vidd is now known as vidd-away04:35
*** rsampaio has quit IRC04:35
troyavidd, are you there ?04:35
vidd-awayim off to bed04:36
vidd-awaywhats up04:36
*** dwcramer has quit IRC04:36
troyai failed to create network with error message >> CRITICAL nova [-] --bridge_interface is required to create a network.04:36
vidd-awayyou must be using flat networking04:37
vidd-awaywhat did you set your bridge to in nova.conf?04:37
vishyuse --bridge_interface04:37
vishyor set --flat_network_interface in nova.conf04:37
troyano, if default it's vlan networking,right ?04:37
troyaor vlan is not recomendded ?04:38
vishyor --vlan_interface if using vlan04:38
troyawhat's better vlan or flat dhcp ?04:39
vidd-awayit all depends on your situation04:39
*** sdake has quit IRC04:39
troyai'm use single node server with one eth004:39
vidd-awaythen you have the same thing i have04:39
troyahow about it ?04:39
vidd-awayim using flat dhcp04:39 off to bed...i have a meeting in the am04:41
troyahave a nice dream vid :)04:41
*** jj0hns0n has quit IRC04:42
*** dendro-afk is now known as dendrobates04:46
*** rsampaio has joined #openstack04:46
*** miclorb_ has quit IRC04:47
*** supriya has joined #openstack04:47
*** miclorb_ has joined #openstack04:47
*** jj0hns0n has joined #openstack04:48
*** lzyeval has quit IRC04:52
*** lzyeval has joined #openstack04:53
*** dysinger has joined #openstack04:58
*** rsampaio has quit IRC04:59
*** supriya has quit IRC05:00
*** dysinger has quit IRC05:02
*** rsampaio has joined #openstack05:24
*** koolhead17 has joined #openstack05:30
*** koolhead17 has joined #openstack05:31
*** rsampaio has quit IRC05:34
pradeeptrying to setup openstack cloud on SLES 11.  fail to sync DB.  " nova sync db"   failed.05:35
*** osier has quit IRC05:36
pradeepany one trying on SLES ?05:36
*** shang has joined #openstack05:37
*** yogi_ has joined #openstack05:41
*** jkyle has quit IRC05:49
*** troya has quit IRC05:55
*** HugoKuo_ has joined #openstack05:56
*** hugokuo has quit IRC05:59
*** localhost has quit IRC06:00
*** localhost has joined #openstack06:01
*** MarkAtwood has joined #openstack06:14
*** chomping has quit IRC06:17
*** jj0hns0n has quit IRC06:18
winston-dhi, guys06:20
winston-dhow can I modify quota for certain user?06:20
*** troya has joined #openstack06:26
*** lzyeval has quit IRC06:26
*** lzyeval has joined #openstack06:27
*** yogi_ has quit IRC06:30
livemoonwinston-d: dashboard or nova-manage project can do that06:30
winston-dnova-manage project or nova-manage account?06:31
lzyevalnova-manage project quota --project=swscloud --key=floating_ips --value=3006:31
lzyevalsomething like this06:31
*** ejat has quit IRC06:31
*** miclorb_ has quit IRC06:32
winston-dlzyeval : thx.  and I found nova-mange account quota --project is the same as nova-mange project quota --project.06:32
winston-dthey both update 'quotas' table in nova database.06:32
winston-dbut still i cann't create more than 10 instances, even after i increase the quota to 1024 and restart nova-api services;06:33
winston-ddo you have any idea?06:33
livemoonwinston-d: default quota you can set in nova.conf06:34
winston-dlivemoon : i see. thx.06:34
*** lmh has joined #openstack06:35
*** carter has joined #openstack06:42
carteranyone here?06:43
carteri am new here, i want to confirm that someone could see my words06:44
*** halfss has joined #openstack06:44
halfsshi,is there anybody know how to use libvirt get instance's disk status?06:45
carteri dont06:49
*** mdupont has joined #openstack06:57
carterin openstack swift, what does partition mean??06:58
*** kaigan_ has joined #openstack07:01
carterin the swift administrator manual, they say "The Ring maintains this mapping using zones, devices, partitions, and replicas. ..."07:02
carterwhat does partions represent?07:02
*** datajerk has quit IRC07:04
*** GheRivero has quit IRC07:08
troyahi all07:11
*** binarychef has quit IRC07:12
troyahi carter07:12
*** nerens has joined #openstack07:13
carterhi troya07:14
troyacarter, i want to create custom image for openstack. i using reference for it07:14
troyathen on command " kvm-img create -f raw server.img 5G " i'm still confuse about this command, carter07:15
*** osier has joined #openstack07:15
troyacarter,can you explains me more about that command ?07:15
troyahow about it carter?07:18
*** ejat has joined #openstack07:21
*** aryan has quit IRC07:24
*** guigui has joined #openstack07:24
cartertroya, my understanding is that this is just a command to create space07:25
troyaso it is not create storage too ?07:26
troyastorage for instances which running use that image ?is it wrong ?07:26
*** mgoldmann has joined #openstack07:29
carteri think it will assign storage07:29
*** mcclurmc has quit IRC07:31
*** mcclurmc has joined #openstack07:31
cartertroya, how do you think?07:32
troyai think that which that command, openstack will create storage which used for  instances which use that image for instances, but i doubt07:33
troyaif my opinion right, then what's different it with nova-volume ?07:34
troyait's which make me litte confuse07:34
*** krow has quit IRC07:38
*** fabiand__ has joined #openstack07:40
*** troya has quit IRC07:41
*** troya has joined #openstack07:41
*** datajerk has joined #openstack07:41
*** siwos has joined #openstack07:44
troyahi all07:45
*** mcclurmc has quit IRC07:47
*** mcclurmc has joined #openstack07:47
*** woleium has quit IRC07:47
*** MarkAt2od has joined #openstack07:48
*** MarkAtwood has quit IRC07:51
*** nRy has quit IRC07:57
*** MarkAt2od has left #openstack08:02
winston-dlivemoon : around?08:09
*** rnorwood has joined #openstack08:12
*** fabiand__ has quit IRC08:21
*** rnorwood has quit IRC08:21
*** dnjaramba has joined #openstack08:25
*** Razique has joined #openstack08:25
*** reidrac has joined #openstack08:25
*** TheOsprey has joined #openstack08:27
*** alekibango has quit IRC08:30
*** alekibango has joined #openstack08:30
Raziquehi all :)08:30
*** dysinger has joined #openstack08:33
*** jedi4ever has joined #openstack08:33
koolhead17hi Razique08:33
Raziquewhat's up dude ?08:34
*** jedi4ever has joined #openstack08:34
koolhead17notthing much, lazy monday08:34
*** jedi4ever has quit IRC08:34
Raziqueenjoyed ur week-end ?08:34
*** jedi4ever has joined #openstack08:34
*** javiF has joined #openstack08:40
Raziquekoolhead17: so, have updated the doc bugs ? :p08:42
koolhead17Razique, its in todays menu!! LD08:42
koolhead17hello livemoon08:43
Raziquelivemoon: :)08:44
troyaWb Razique :)08:46
Raziquetroya:  :)08:49
*** foexle has joined #openstack08:50
Raziquewhat's up ?08:50
*** pixelbeat has joined #openstack08:50
troyaRazique: now i trying to create custom image with ubuntu 11.04, i following reference from
*** HugoKuo__ has joined #openstack08:50
troyathen when i try command >> sudo kvm -m 256 -cdrom ubuntu-11.04-server-amd64.iso -drive   file=server.img,if=scsi,index=0 -boot d -net nic -net user -nographic  -vnc :008:51
*** guigui has quit IRC08:51
troyaerror message like this >>
troyait's appear that i need custom ip addrss08:51
Raziquetroya: replace vnc:0 by -vnc:1008:51
*** dysinger has quit IRC08:52
*** nacx has joined #openstack08:52
troyai have replace it, but console output still error >> kvm: pci_add_option_rom: failed to find romfile "pxe-rtl8139.bin" and then blank on below of it08:53
*** HugoKuo_ has quit IRC08:53
*** dysinger has joined #openstack08:54
Raziqueok that means u need that drivers08:55
Raziqueit's no biggie, u can ignore it08:55
Raziquedoes kvm start anyway ?08:55
Raziqueif no, l've the drivers, u awnt em ?08:55
troyastill blank cursor08:55
troyarazique: thanks, how i get it from you ?08:56
Raziqueblank cursor -> means the command stopped, or is it sill running ?08:56
Raziquetroya: mmmm can I ssh to ur server, so I scp them ?08:56
*** miclorb_ has joined #openstack08:56
troyastill running, but only blank08:56
Raziqueof do u have a ftp ?08:56
Raziquetroya: ok that means u can connect :)08:56
Raziqueit's a bug in ubuntu pacakge, drivers are missing08:56
troyacan or cannot ?08:56
troyaok, so i must get driver first08:57
troyaso it's success ?08:57
Raziquewell, you can add them, but still works without them08:57
Raziqueyup, it the command is running (means u haven't get the shell back)08:58
Raziqueu can VNC to the kvm instances08:58
troyayees, i get it. but what's username and password to login into ?09:00
troyaRazique,what's default username and password, i have tried use my local system password but failed09:06
Raziqueno user/ no password :)09:08
Raziqueand port 5910 :)09:08
Raziquevnc:0 : 590009:08
Raziquevnc:10 : 591009:08
Raziquevnc20:  5920 etc...09:08
troyavncviewer, right09:08
troyait's need login on me09:09
*** guigui has joined #openstack09:10
Raziquetroya: no u don't need any login09:10
Raziquewhat the client u are using ?09:11
*** SpamapS has quit IRC09:11
*** SpamapS has joined #openstack09:12
troyai work on ubuntu 11.1009:12
Raziquehow do u connect to vnc ?09:12
troya2.55 is my ip09:13
troyabut now when i execute it error like >>  main:        unable to connect to host: Connection refused (111)09:13
Raziquethe port is the wrong one :)09:13
troyahehe,,,,it's appear now09:14
*** dachary has joined #openstack09:14
troyaubuntu installation :)09:14
troyayou are cool ")09:14
Raziquegreat :)09:15
Raziquedon't hesitate if u have any question09:15
koolhead17troya, :)09:15
troya i have question again, firstly i execute command : kvm-img create -f raw server.img 5G09:16
troyais it mean that i create file system of my new image with size 5Gb ?09:16
Raziquebut I would advise qcow2 instead of raw09:17
Raziquethat file will be seen as the hard drive for the image09:17
troyawhy ?09:17
Raziquetroya: because u'll be able to do thin-provisionning09:17
Raziqueeg qcow of 50gb -> the image sees 50gb09:17
Raziquebut if it only uses 2 gb, then the image into ur store will weight 2 gb09:18
RaziqueMaybe the raw allows this also, but haven't tested ^^09:18
troyarazique: what's you mean eg,09:19
Raziqueu'll be able to create a file the size u want09:21
Raziquelet's say 50 gb kvm-img create -f qcow2 server.img 50G09:21
Raziquenow during the install, ubuntu will see 50 gb09:21
Raziqueat the end of the installation, it the system occupies 3 gb09:22
Raziquethe file itself won't weight 50gb but 3 gb09:22
*** jedi4ever has quit IRC09:22
Raziqueinto the glance repo, that means u gain 47 gb :)09:22
*** jedi4ever has joined #openstack09:23 system only take 3 Gb and the others (47 GB) for storage, right ?09:23
*** dobber has joined #openstack09:23
Raziquetroya: no, that's not what I mean ^^09:24
Raziqueactually u are creating an ubuntu image09:24
Raziqueu assign it a disk via kvm-img create $disk09:25
Raziquethat disk is the one your image will see09:25
Raziqueby using qcow2 you can give it a bigger size09:25
troyayes, i will see it on my installation procedure09:25
*** foexle has quit IRC09:25
Raziquefor instance 50gb09:25
Raziquebut on the host, the size will be the real one09:25
Raziquethe effective one09:25
Raziqueat the end of ur ubuntu installation, u won't use 50gb, but most commonly around 3gb09:26
Raziqueso the file u create with kvm-img create will only weight 3 gb09:26
Raziquenot 50 gb09:26
troyaso hm..what different if i use raw with size 50 G ? is it same ? i can assign partition when installation procedure to determine between /, /home or the others ?09:28
*** rustam has joined #openstack09:28
*** foexle has joined #openstack09:29
Raziquetroya: with raw, i don't know if the size will be dynamic09:29
Raziquemaybe with raw, u'll end up with a 50 gb file09:30
Raziqueeven if the ubuntu install only used 3 gb09:30
*** jedi4ever has quit IRC09:31
troyamay be will be clearly if i try directly :D09:31
troyaRazique: when i execute command : " kvm-img create " to create storage.Is Size of my file system will be reduce too ?09:38
*** redconnection has joined #openstack09:38
*** jedi4ever has joined #openstack09:38
Raziquenope that command only creates an empty file09:39
Raziquebut if you install an OS on it, yes it'll grow09:39
troyathen how about network configuration ? if on installation i choose DHCP, is it will get DHCP from controller or from router on my network ?09:40
*** scottjg has quit IRC09:49
*** littleidea has joined #openstack09:51
*** bourke has joined #openstack09:54
Raziquetroya: for the install u can choose the conf u want09:55
Raziqueat the end, don't forget to use DHCP09:55
Raziquethe dnsmasq proces (managed by nova-network) will send an ip adress09:57
troyayes, it have finished, now i upload it to nova09:58
doude_Hi all, some one use the OpenStack API extension 'flavor extra spec' ?09:58
troyaRazique: after i execute this command >> sudo kvm -m 256 -drive file=server.img,if=scsi,index=0,boot=on -boot c -net nic -net user -nographic -vnc :0 , so must i connect with vncviewer again ?09:58
*** Zhann has quit IRC09:59
Raziquetroya: u used raw ?10:00
*** Zhann has joined #openstack10:00
troyayes,  i have not to change to qcow yet10:01
Raziqueok ok u use scp for transfer ?10:03
livemoonI will go home10:04
*** livemoon has left #openstack10:05
troyarazique :ubuntu driver do you mean ?10:05
troyalivemoon: bye10:06
*** darraghb has joined #openstack10:12
troyarazique: what's the meaning 1048576 on  >> losetup -f -o 1048576 server.img10:14
*** GheRivero has joined #openstack10:16
*** jedi4ever has quit IRC10:22
*** foexle has quit IRC10:23
*** jedi4ever has joined #openstack10:27
*** lzyeval has quit IRC10:29
Raziquetroya: it's used if u want to extract the data partition10:29
troyathen 1048576, what size is it ?10:30
Raziqueit's the end of the data partition10:30
Raziquethat value depends on ur install10:30
Raziquedon't use the same as the doc10:30
troyawhere i can get that number ,10:30
troyafrom fdiks ?10:31
troyaor any other rules?10:31
Raziquemm the guide explain before as far as I recall ?10:33
troyao..ya..i get it10:35
*** foexle has joined #openstack10:45
*** coli has joined #openstack10:53
*** junaid has joined #openstack10:53
*** ug has joined #openstack10:55
ugcan Keystone be configured for using syslog ?10:55
*** troya has quit IRC10:58
foexlehey guys, why shows me "euca-describe-instances" all instances in running state although many instances are shut off ?11:00
foexlei'ts after a reboot of compute-node11:00
foexlenormally should the compute node update the database or ?11:01
*** ahasenack has joined #openstack11:02
*** miclorb_ has quit IRC11:04
*** ug has quit IRC11:04
Raziquefoexle: hehe man :)11:05
Raziquenope, after a node reboot, the database ain't updated11:05
Raziquei wrote somehting about this here
*** ejat has quit IRC11:09
*** guigui has quit IRC11:10
*** DuncanT has joined #openstack11:10
*** pothos has quit IRC11:11
foexleRazique: hey Razique :) how are you ? :) .... so i read your doc, but this are not practicable in big clouds. Only a question: why don't destroy any relationships for all vms and other stuff like ip's, volumes ... for this compute node automaticly ?11:14
foexlethis vms are stateless, so i have a clean db after a crash and can spawn simply new vms11:15
Raziquefoexle: how are you suppose to know when the node goes down ?11:16
Raziquethat explains why the db ain't updated11:16
*** guigui has joined #openstack11:17
Raziquethat is linked to nova HA mecanisms that we are debating here for weeks11:17
Raziquetoday nova clearly lacks one some HA parts11:17
Raziquefor instance, today if a node goes down, it's a nightmare11:17
Raziquenova should be able by itselft to power on the instance elsewhere11:17
*** ug has joined #openstack11:18
ughow to configure Keystone to use syslog11:18
Raziqueug: into keysone.conf11:18
Raziqueuse_sylog : true11:18
Raziquethen restart Keystone11:19
ugI don't see the entry 'use_syslog' in keystone.conf11:19
foexlethis is autohealing ... so yeah would be a nice feature :) .... but again, this will be the second step (autospawning), but the first should be keep the cloud clean without db hacking11:19
foexleor ?11:19
Raziquefoexle: thought if u have any idea to automatically mograte instance when the node goes down, then let me know :)11:19
Raziquefoexle: can't happen today11:19
ugRazique: you mean I will need to add the entry use_syslog ?11:19
Raziquenova doesn't update the database11:19
*** littleidea has quit IRC11:20
Raziqueug: no, let me find out the setting11:20
ugRazique: ok11:21
Raziqueug: you are right11:23
RaziqueI mixed with other projects conf11:23
foexleRazique: this can be a indicator to clean your database after a disaster (Found 4 in the database and 1 on the hypervisor.)11:23
foexlefound this in the log on the compute node11:23
Raziquefoexle: not really, since u can have an instance spawning11:24
Raziquewhen an isntance is being started, the instance has an entry into the DB11:24
Raziquebut not running yet11:24
Raziqueif so, you have that message11:24
ugRazique: I updated the 'handlers' in etc/logging.cnf to 'production' only and restarted Keystone11:25
*** jedi4ever has quit IRC11:25
*** PotHix has joined #openstack11:25
ugbut the start up logs still go into keystone.log11:25
Raziqueug: maybe syslog is not supported yet :11:26
ugok, thanks Razique11:26
Raziquei'll check the commits11:26
foexlebut i think, cloud = instance not running => don't have DB entry, because all instances are stateless and i can spawn a new one. So if a instance in db but not running, delete in db. Or compute and network reads from db all attributes and this services recover it self. Like spawn all instances they are in db and make network settings11:27
Raziquemmm nothing about syslog11:27
*** dysinger1 has joined #openstack11:28
foexleor maybe a CLI command or somethink like that11:28
foexlebut it should be automated11:28
Raziquefoexle: complicated.... because you suppose instances have two states only 1 : running 2 : not running11:29
Raziquein reality, an instance could be resizing, migrating, powered off, paused etc...11:29
*** dysinger1 has quit IRC11:29
ugRazique: thanks11:29
*** guigui has quit IRC11:29
Raziqueso that justifies a DB entry just after you request a spawning11:29
*** dysinger has quit IRC11:29
Raziquefoexle: plus, suppose you loose connectivity to the node for let's say 5 mn11:30
Raziquewhat nova should do, update the database, remove the entry ? that means when the node recovers connectivity, u end up with a running instance, but unreferenced :)11:31
*** junaid1 has joined #openstack11:31
Raziqueanother use case would be the network part : nova-network uses the database in order to create the routing rules for the instances11:31
*** junaid has quit IRC11:31
foexleRazique: yeah i know :D ... i think you dont understand what i mean ;) .... so node goes down => node comes up => node read all attributes from database (instances, ips, volumes etc.) => node will autoheal it self like spawning instances, making network, attach volumes and ...11:31
Raziqueremoving the entry doesn't remove the current iptables rules of the instance :/11:31
RaziqueYah, that 's the ideal behaviour that doesn't happen today11:33
Raziquethus the all-by-hand hack of mine =D11:33
foexleahhhhhhhhhhhhh ok :D .... so this point are on the roadmap11:33
RaziqueI don't know... doesn't look so11:34
Raziquehehe, that is my biggest fear man todayt11:34
Raziqueif a node goes down, it's a real noghtmare11:34
Raziquebecause the "controller" doesn't do anything11:35
Raziquethe disk are still marked as attached, instance as running11:35
Raziqueeven worse.... I've seen one day after a node failure11:35
Raziquethat nova decided to remove all files that belong to the instance (/var/lib/nova/instance/instanceXXX) since it was not running11:36
*** livemoon has joined #openstack11:36
foexlehmmm yeah why not ? ... so you can make a CLI command like nova-manage service clean <ID> to delete all db entries of this service (only instance, IP, volumes) and tell this services delete all files11:39
foexleits a dirty workaround11:40
foexlebut you dont get inconsistency in your db11:40
*** katkee has joined #openstack11:40
*** frankon has joined #openstack11:42
*** ug has quit IRC11:42
Raziquefoexle: that should do it11:44
*** ZX-14 has joined #openstack11:44
Raziqueif u run it manually yah11:44
*** ZX-14 has quit IRC11:44
Raziquebut automatically requires a bit of reflexion11:44
*** cereal_bars has joined #openstack11:44
Raziquewhat would be the criterias for such case11:44
RaziqueI've a node on which run 40 instances, If I loose connectivity, should I remove everything ?11:45
Raziqueof course not...11:45
Raziquethat is why it's a bit complicated11:45
foexleif you get a timeout > "conf settings" => yes11:45
foexlethe second step was to autrespawn this instances on a other host11:46
Raziquewouldn't be the first step ?11:46
Razique1 : I losse connectivity, instances respawn on a connected node11:47
Razique2- remove everything from the second node11:47
Razique3- now suppose the server is still up (the loste node)11:47
Raziquewhat happens when it regain connectivity ? i'll still have the instances running on it11:48
Razique4- another thing to consider is the data, in such case you loose the data (that happened to me :D )11:49
foexle1: loose connectivity, tell "recover service" => I loose connection of this node, please repair11:49
foexle2: this services looks into the db, which ip's are using, volumes and all informations11:50
foexle3: this services tells nova-api spawn instance xxx (same attributes as the old one)11:50
*** Hakon|mbp has joined #openstack11:51
foexle4: setting a new "Flag" in the DB which indenticates this is a broken instance => after starting instace on a new node setting this flag to True11:51
foexle5: other compute node is up again11:51
foexle6: compute node tells DB which instances i have11:52
foexle7: compute node see => ok all my instances, route settings volumes etc. will be broken => destroy all11:52
Raziquefoexle: what about the data ?11:52
foexle8: you have auto healing and a clean node if they comes up again11:52
foexlewhich data ?11:53
Raziquecontained inside the running instainces11:53
Raziquethe files u put etc...11:53
foexleit's a cloud, it's stateless you can use blockdevice like iscsi or swift to upload datas. But all datas on each vm disk should be stateless11:54
Raziquethat the things, it's never really stateless11:54
Raziqueusers u have created, packages you have installed etc...11:54
foexlethats not your problem11:54
Raziquehostname of the server11:54
foexlefor this tasks do you can use chef11:55
foexleor puppet11:55
foexleor complete images11:55
foexleif you look at ec2 you have vm disk space, but this is completly stateless11:55
foexleif you need a storage11:56
foexleyou can attach iscsi or use swift11:56
*** stevegjacobs_ has quit IRC11:56
*** osier has quit IRC11:58
Raziquefoexle: need to luch11:59
Raziquei'll think about that while eating11:59
foexleRazique: ok :) good hunger11:59
Raziquethat 's a very intereting debate11:59
Raziquedunno today into which measure is it possible to have isntances stateless12:00
Raziquemy customers today clearly don' t have stateless ones12:00
Raziquecya later foexle :)12:00
foexlecu Razique :)12:00
*** javiF has quit IRC12:03
*** stevegjacobs_ has joined #openstack12:07
*** dnjaramba has quit IRC12:12
*** dnjaramba has joined #openstack12:13
*** ejat has joined #openstack12:13
*** ejat has joined #openstack12:13
*** Hakon|mbp has quit IRC12:14
*** Eyk^off is now known as Eyk12:14
*** redconnection has quit IRC12:17
*** redconnection has joined #openstack12:18
*** abecc has joined #openstack12:21
*** redconnection has quit IRC12:22
*** bsza has joined #openstack12:31
*** junaid1 has quit IRC12:43
*** dnjaramba has quit IRC12:46
*** dnjaramba has joined #openstack12:46
smoserKiall, there was a desktop version of, but it did not end up going live. only to a beta at one point.12:47
Eyk (Swift) I have created rings with a replication of 2 and want to increase it to 3. Can I simply recreate the ring without losing any data?12:48
*** javiF has joined #openstack12:49
* koolhead17 leaves .12:49
*** halfss has quit IRC12:50
*** pradeep has quit IRC12:50
*** dysinger has joined #openstack12:51
*** dwcramer has joined #openstack12:52
*** dysinger has quit IRC12:56
*** koolhead17 has quit IRC12:56
*** vernhart has quit IRC13:03
*** jedi4ever has joined #openstack13:04
*** Eyk is now known as Eyk^off13:05
*** jedi4ever has quit IRC13:05
*** dwcramer has quit IRC13:07
*** nerdstein has joined #openstack13:09
notmynameEyk^off: yes and no. if you completely change your ring like that, you won't lose any data, but it will all become unavailable (because the data won't be rehashed). when you chnage the ring like that (same thing if you change the partition power) all of the data must be migrated (ie GET(old)+PUT(new)). it can be done, but it will take your own dev work13:11
*** javiF has quit IRC13:12
notmynameEyk^off: perhaps one possibility would be to manually edit the ring file to remove one of the replicas. I'm not sure how well that would work, though, because that doesn't change the builder file (which it required for ring updates)13:12
notmynameEyk^off: basically, you could probably reverse engineer it (and I'd support that). but the simple answer is "no, don't do that"13:13
*** redconnection has joined #openstack13:14
foexledo anyone have a problem with attaching ssh keypair ?
foexlestart the instance normally with -k <keypair-name>13:19
foexlepriv_key.pem are genrated by openstack13:19
*** guigui1 has joined #openstack13:20
*** praefect has joined #openstack13:33
*** chomping has joined #openstack13:34
*** jedi4ever has joined #openstack13:40
*** bonzay is now known as zz_bonzay13:43
*** zz_bonzay is now known as bonzay13:44
*** osier has joined #openstack13:44
*** jedi4ever has joined #openstack13:47
*** mattray has joined #openstack13:49
*** MarkAtwood has joined #openstack13:50
*** sdake has joined #openstack13:50
*** livemoon has left #openstack13:50
*** PeteDaGuru has joined #openstack13:51
*** lts has joined #openstack13:53
*** GheRivero is now known as GheAway13:55
*** mrjazzcat has joined #openstack13:57
*** mrjazzcat has left #openstack13:57
*** marrusl has joined #openstack14:01
*** dgags has joined #openstack14:02
*** markvoelker has joined #openstack14:04
*** martine has joined #openstack14:05
*** martine has quit IRC14:07
*** martine has joined #openstack14:07
*** lorin1 has joined #openstack14:09
*** Shentonfreude has joined #openstack14:10
*** vernhart has joined #openstack14:12
foexleRazique: wb14:14
*** hadrian has joined #openstack14:15
Raziquefoexle: ok ok u literally make me thing during my whole lunch time :D14:16
foexlei'm sorry ;)=14:17
RaziqueI think ur point of view is excellent14:17
Raziquebut kinda hard to apply14:17
Raziqueok let's say a customer install a new application14:17
Raziquenow ur node crashes, u respawn, puppet etc...14:18
Raziquehow do you restore his files ?14:18
foexleyou need to ask you, which files need this user14:18
foexleyou have application files14:19
foexlelike images14:19
foexleor attachments or other files14:19
foexlethis files can you handle wiht swift14:19
foexleyou have log or db files14:19
foexlethen you need block storage => glance14:19
foexleso you dont need any persistent storage on each vm14:20
foexleso lets focus on instance types14:21
Raziqueglance doesn't manage block storage it's just a delivery service14:21
Raziqueu mean nova-volume14:21
foexleyeh iscsi14:21
foexlei'm sorry14:21
Raziquewhat about configuration let's say /etc/apache14:21
foexleor puppet14:21
foexlewe do all this things with chef14:22
foexleund we distinguish between application instances, utility intances and db instances for each project14:22
foexleso we know14:22
foexlethis instances goes down in this security group14:23
foexleit was a application instace maybe14:23
foexleopenstack "autoheal" an spawning a new instance in the same security group14:23
Raziqueok, now if my customer updates an apache vhost14:23
Raziquesomething that I wouldn't be aware of14:24
foexleso we can handle this with chef "please load role for application instace by project"14:24
foexleyeah a good example14:24
foexleyou customer say to you .... please change my vhost14:24
foexleor add a new application on the same hosts14:24
foexleyou need only one time to change this todo on your chef server14:25
*** lorin1 has quit IRC14:26
foexleand all running instances with this role or recipe will change this settings14:26
foexleand new instances too14:27
Raziquefoexle: that the thing14:27
*** lorin1 has joined #openstack14:27
Raziquecustomers don't report that :d14:27
Raziquethey create users, update configurations14:27
Raziqueinstall stuff in /opt14:27
Raziqueso in fact sometimes you are not aware when the server files evolve14:28
foexlebut this are not the normal cloud domputing logic14:28
foexleif you test, ec2, hp-cloud, vm sphere you will see every instance are stateless14:28
foexleyou can mapp storage devices or put some files to S3(swift)14:29
foexlebut all instances are stateless14:29
Raziquefoexle: on a pure "logical" approach I 100% agree with you :)14:30
Raziquebut on a more "pragmatical" way14:30
*** lorin1 has quit IRC14:30
Raziqueyou have production, a node crashes, the customer doesn't care about the "state of art about cloud"14:31
Raziqueyou see what I mean ?14:31
*** vernhart has quit IRC14:31
Raziquemy priority is to make sure I'm quickly able to restore data, if i can't, I'm not allowed to simply say "You should have told me about that new directory that contains hyper important stuff"14:31
foexleyeah but then need this customer support .... you can not build a fucking good tool set for stupid or ignorant customers14:32
Raziquefoexle: hehe good point here but you are as aware as I am that in real life, they sometimes can be stupid14:32
Raziquethat don't make them no customer14:32
Raziquethat is why I was thinking into which measure I could turn these instance as stateless as possible14:33
Raziquemaybe having a script that crawls the instance and reports the changes ?14:33
foexleRazique: i know ... i know :) .... but with this way, if you say a customer can save datas in each vm and this datas cant be lost, you can't realize a autohealing or autoscaling and this two points are one of the unique selling points of ec214:34
Raziquefoexle: I think Openstack shouldn't be considered as an opensource ec2 cloud14:35
Raziqueit has more potential14:35
Raziquefoexle: here we are trying to have the flexibility of ec214:35
Raziqueand features of vmware :d14:35
foexle:D yeah thats nice i hope so :D14:35
Raziquefoexle: regarding the datas, I already use nova-volume14:36
Raziquefor mysql, apache, etc...14:36
*** CatKiller has quit IRC14:36
Raziqueopenstack has implemented a "suspend" mode14:36
Raziquethat allows you to power off the instance14:37
foexlethats the correct way .... if the instance craches, fireing up a new instance and map the existing volume to the new one14:37
Raziquefoexle: one day I've lost an instance14:37
foexleand you have all persistent datas in each vm14:37
RaziqueI recreated it, reattached the volume14:37
Raziquetwo things took me time14:37
foexleyes this should be automatically14:38
Razique1- Reconfigure the server, and 2- explain to all customer that the SSH id has changed14:38
foexle1: Instance goes down14:39
*** jedi4ever has joined #openstack14:39
foexle2: new instance will spawn14:39
foexle3. attach volume14:39
foexle4: starting chef with correct role14:39
*** swill has joined #openstack14:39
foexle5: chef are done, automatically send email to customer list14:39
foexle3.1 attach correct pub ip14:40
Raziquefoexle: have u written a script for that ?14:40
Raziqueif not I could try one14:40
foexlewe do that with chef14:40
Raziquesomehting you would cron, that would check the nodes regularly14:41
Raziqueno I mean the node failure detection14:41
Raziquedoes chef has recipes for opentack ?14:41
*** rods has joined #openstack14:41
foexleso i dont run openstack in prod .... i'm in testing phase :) .... this described steps are on ec214:42
Raziqueah ok14:42
foexlei think amazon have a service and check the insances with keepalived14:42
foexleand if a instance goes down this services grap all infromation of the "broken" instance and fire up a new one14:43
Raziquefoexle: I'll try to script something similar14:43
RaziqueI've used GlusterFS in order to sync. instances accross nodes14:43
Raziqueso if a node goes down, you simply reboot the instance on another node14:43
*** carter has quit IRC14:44
Raziquethus you don't losse data14:44
*** reidrac has quit IRC14:44
*** mdomsch has joined #openstack14:44
*** stuntmachine has joined #openstack14:44
*** MarkAtwood has quit IRC14:44
foexlebut you have more net traffic with permanently syncing14:44
Raziquefoexle: the overhead is not that big14:44
phschwartzHey all, I am closer to having a working environment. I can spawn nodes, but when doing the following action. waiting for metadata service at The intances timeout 100 times then fail to setup keys.14:44
phschwartzany idea?14:44
Raziquemoreover a networks that doesn't work doesn't exist =D14:44
foexlebut useless if you have stateless instances14:45
Raziquefoexle: totally14:45
foexleand you are more flexible14:45
Raziquefoexle: why is that ?14:45
Raziquephschwartz: check the iptables rules, does a nat exist14:45
Razique169.254:80 redirects to $nova-api-ip:877414:45
foexleso you can install a new compute node in 5 min and you can directly spawn instances on this node14:46
*** nerens has quit IRC14:46
foexlewithout syncing or image transfer14:46
foexleyou use glusterfs14:47
phschwartzRazique: DNAT       tcp  --  anywhere        tcp dpt:www to:
foexlewich fs do you use ?14:47
foexleso yo need gfs/gfs2 or ocfs214:47
foexleto handle simultaneously write access14:48
Raziquefoexle: I use glusterFS14:48
Raziquein fact it's speeder since the files are already present on all nodes14:48
foexlethis is verry susceptible .....14:49
Raziquefor spawning, the image needs to be sent to the compute nnode14:49
*** deshantm_laptop has joined #openstack14:49
Raziquewhat do u mean ?14:49
foexleglusterfs are not a filesystem ... which filesystem do you use in glzusterfs?14:49
Raziquephschwartz: run netstat -tanpu | grep 877414:49
*** siwos has quit IRC14:50
Raziquefoexle: glusterFS is a filesystem actually14:50
Raziquephschwartz: ok is the instance run on another node ?14:50
phschwartzno, it is all a single node14:51
Raziquefoexle: shouldn't we start to think about HA mecanisms a write a doc ?14:51
Raziquestart -> continue :D14:51
Raziquephschwartz: ok can I see the output of "euca-get-console output $instance"14:52
Raziqueor let's continue here, i'll save the log and write a doc after based on what we said14:53
Raziquefoexle: so at Amazon, if you create a file on let's say /opt14:53
Raziqueand amazon node crashes, you loose the file ?14:53
phschwartzRazique: This is an instance setup using Kiall's scripts. I am new to openstack, how can I generate an EC2_ACCESS_KEY. I started the node using dashboard/keystone14:53
Raziquephschwartz: nova-manage project export14:54
foexleRazique: yes14:54
Glaceerazique: if its on  local storage yes.. not on ebs14:54
KiallRazique, nope... doesnt work with keystone14:54
foexleGlacee: :D14:54
RaziqueKiall: oh yes, my bad14:54
Raziquesince it's Keysone which handles the tenants14:55
Kiallphschwartz: you need to add EC2 creds to keystone, from my scripts you can see an example at the end of
KiallNova/Keystone will not generate a key for you... You have to pick them youself and tell keystore what they are14:55
Glaceefoexle: ebs has been so slow lately though..14:55
Raziquefoexle: Glacee i was also thinking about automatiing the creation of images from running instances14:56
phschwartzkiall: if I used your generated data script to initialize it, any I dea what it would be?14:56
Raziquethus if a node fails, I spawn an instance based on that image14:56
Raziquethat mean, in worst case scenarios I loose a day of conf. (since the datas are on the volume)14:57
Kiallphschwartz, my data adds 1 key for the admin user.. Access key = admin_$admin_password and Secret_Key = admin .. if i remember right14:57
foexleif a node fails you can look in your db14:57
foexlether are all informations14:57
foexleof all instances on this node14:57
Kiallphschwartz, or add new ones with something along the lines of`keystone-manage credentials add $username EC2 $access_key $secret_key $tenant`14:58
foexleso you can spawn with all things (volume, pub ip, image type ....) on other instances14:58
phschwartzkiall: ok, that got me further. I did the euca-get-console-output and got the following error: EC2_URL not specified. Trying http://localhost:8773/services/Eucalyptus14:58
foexleRazique: the next good thing if you have stateless instances14:58
Kiallphschwartz, 1 sec.. let me grab the template novarc I use14:59
*** mjmac has joined #openstack14:59
foexleRazique: you can say .... i need every time 5 instances on this project, if i loose one please spawn a new one"14:59
*** mjmac has left #openstack14:59
foexletime to go quickly some a cigarette ;)15:00
Glaceeis there a reason why I cant find the swift-stat system section on the new dev doc version> 1.4.5? is it being removed?15:01
*** lionel has quit IRC15:01
*** lionel has joined #openstack15:02
Kiallbear in mind, that example file uses a different EC2 access/secret key.. I changed the format of mine so there was less settings...15:02
*** bsza has quit IRC15:03
*** chemikadze has quit IRC15:03
foexleok back15:04
phschwartzKiall: I think I got it set right for me. But using the euca command again I get: Warning: failed to parse error message from AWS: <unknown>:1:0: syntax error15:04
*** egant has joined #openstack15:04
Kialltry with --debug (or is it --verbose) and see what the error is15:04
phschwartzhmm, The server could not comply with the request since it is either malformed or otherwise incorrect.15:05
*** martine_ has joined #openstack15:05
foexleKiall: if start a instance with a keypai and attach a public ip, i can't connect with the correct private key
*** martine has quit IRC15:06
foexleKiall: do you know what happens?15:06
*** chemikadze has joined #openstack15:07
*** PiotrSikora has quit IRC15:07
Kiallfoexle, that looks like you are using the wrong type of SSH key... but, I've never seen anything like that...15:08
Kiall(the givaway .. debug2: key_type_from_name: unknown key type '-----BEGIN'"15:08
foexleyeah but its a correct type15:08
foexle-----BEGIN RSA PRIVATE KEY-----15:09
*** PiotrSikora has joined #openstack15:09
*** bcwaldon has joined #openstack15:09
KiallWell, SSH doesnt agree ;)15:10
phschwartzkiall: the error:  my novarc:
foexlei know .... but i dont have change this keypair :( and last week this works well -.-15:10
*** kbringard has joined #openstack15:11
Kiallfoexle, honestly.. It sounds like a valid SSH key, its your local install of SSH rejecting it, not nova or the intance.. Thats all I can tell!15:12
*** dubsquared has joined #openstack15:12
Kiallphschwartz, honestly, I dont know.. I would bet the key or secret is wrong...15:12
KiallAnyway - I've got work to do.. cyas15:12
foexleKiall: my local ssh client ? .... hmmm i try it from 3 machines :( .... ok i'll look15:13
Kiallprobably means its the keyfile so...15:13
Kiallfoexle, did you by any chance pass that key via a windows box?15:14
Kiall(I have no idea oif mucking up the unix line endings would cause an issue, just a guess)15:14
foexleonly linux15:14
KiallAh well - scratch that idea then..15:14
phschwartzRazique: is there any other way to get the console output you want other then with euca2ools? I cant ge them working at the moment and not sure why.15:15
*** nerens has joined #openstack15:15
phschwartzI can get the console log from dashboard15:15
foexlephschwartz: virsh console <domain>15:15
*** kbringard has quit IRC15:15
phschwartzhow can I get the domain?15:17
foexlevirsh list --all15:17
phschwartzRazique: the console reports error: internal error character device (null) is not using a PTY15:17
*** CaptTofu has joined #openstack15:19
*** Razique has quit IRC15:19
_rfzHi guys, I'm using FlatDHCP network and I can't ping or access my VM's - i've added the euca-authorize -P icmp -t -1:-1 default  and ssh15:19
*** Razique has joined #openstack15:19
foexle_rfz: iptables -L -t nat15:19
foexle_rfz: look if the roules are set correctly15:19
Raziquefoexle: ok back15:20
Raziquecoffee for me, I'm no smoker :p15:20
foexleRazique: :D wb15:20
foexleRazique: do you have a soloution of my problem ?15:20
Raziquewhich one ? :D15:21
Raziquesorry I've missed something hjere :D15:21
foexleif start a instance with a keypai and attach a public ip, i can't connect with the correct private key
_rfzfoexle - they rules seem to fine15:21
foexleeven ssh key i use the same issue15:21
foexle_rfz: can you ping from host ?15:22
Raziquefoexle: display the file15:22
Raziqueis it a private key ?15:22
foexleRazique: yeah15:22
foexle-----BEGIN RSA PRIVATE KEY-----15:22
*** j^2 has quit IRC15:23
*** dendrobates is now known as dendro-afk15:23
_rfzfoexle I can't ping or access the VM on the public IP - I can ssh and ping on the internal IP15:23
foexle_rfz: do you have assotiated the pub ip to the vm ?15:24
Raziquefoexle: telnet the instance on port 2215:24
*** j^2 has joined #openstack15:24
Raziquesounds stupid but we never know15:24
foexleRazique: telnet output SSH-2.0-OpenSSH_5.8p1 Debian-7ubuntu115:24
foexleyeah i have seen this never before15:25
_rfzfoexle yep, it has the public IP assigned to the VM15:25
foexle_rfz: i cant see a public address15:25
*** chemikadze has quit IRC15:25
foexle192.168 are reserved for local networks15:26
foexlein w3c standard15:26
Raziquefoexle: how did u generate the keypair ?15:26
foexleRazique: thats the same keypair that you have generated :D15:26
*** hallyn_ has joined #openstack15:26
_rfzfoexle so I need to use a real publlic IP range?15:26
Raziquefoexle: ah ok via nova-manage project export then15:26
foexleRazique: yes15:27
Raziquefoexle: which image are u using, the one we retrieved ?15:27
*** dendro-afk is now known as dendrobates15:27
foexleRazique: yes all the same15:27
foexleRazique: i have only change the server ips ... but that can't be the problem15:27
Raziquenope, not really15:27
*** kbringard has joined #openstack15:27
foexle_rfz: if you login from extern yes ... if you try it from the same box not15:28
*** krow has joined #openstack15:28
_rfzfoexle - let me change it :) thanks15:28
Raziquefoexle: regenerate a new set of keys15:29
*** ejat has quit IRC15:30
praefecthi everybody, anyone has a good pointer to setup a multi-zone environment? I don't even know where to start, can I have 2 zones with a single EC2 API url?15:30
*** ejat has joined #openstack15:31
*** ejat has joined #openstack15:31
foexleRazique: yeah :D i'm looking for the command ;) wait15:32
*** bourke has quit IRC15:32
foexlecan i get the pem of existing keypair ?15:33
*** bourke has joined #openstack15:33
*** ejat has quit IRC15:33
Raziquefoexle: not sure15:34
*** keekz_ is now known as keekz15:36
foexleRazique: ok runs15:38
*** lorin1 has joined #openstack15:38
foexleRazique: so i think the failure was, the server have not generated a server key15:38
foexleRazique: i mean this one -----BEGIN SSH HOST KEY FINGERPRINTS----- ... this output was not by the other key15:39
foexleinstance output15:39
*** hallyn_ has quit IRC15:40
*** dprince has joined #openstack15:40
Raziquefoexle: I've already seen that15:43
foexle * Starting execute cloud user/final scripts                             [ OK ]15:44
foexlethis is the last line15:44
foexleby the other key15:44
*** ldlework has joined #openstack15:44
*** sticky has quit IRC15:45
Raziquesounds good15:45
*** code_franco has joined #openstack15:46
*** jeremy has joined #openstack15:49
*** cereal_bars has quit IRC15:50
*** javiF has joined #openstack15:52
*** rnirmal has joined #openstack15:53
*** primeministerp has joined #openstack15:54
*** osier has quit IRC15:54
*** sticky has joined #openstack15:55
*** andrewbogott has joined #openstack15:55
*** adjohn has joined #openstack15:56
*** jkyle has joined #openstack15:58
kbringardhey Razique15:59
Raziquehey kbringard15:59
Raziquehow are you ?15:59
kbringardI'm OK16:00
Raziqueenjoyed ur week end ? :)16:00
*** dragondm has joined #openstack16:00
kbringardheh, I've had better :-)16:00
kbringarddad had a heart attack, so I ended up driving 1400 miles from Colorado to California16:00
*** jkyle has quit IRC16:00
kbringardhe's fine, but now I'm friggin tired16:00
kbringardbut, at least I get to spend Thanksgiving with the family :-)16:01
kbringardhey, if you have a moment, nerdstein is having an error when he tries to euca-describe-images16:02
Raziquewhat's the output ?16:02
*** TheOsprey has quit IRC16:03
annegentlekbringard: oh so sorry to hear it. That's a long drive too. Glad to hear he's ok, hope you can catch up on sleep this week.16:03
Raziquehey annegentle :)16:04
kbringardannegentle: thanks16:04
kbringardit's even longer with a 3 year old and a 1 year old in the car :-D16:04
annegentlekbringard: oh yeah believe me I know :)16:04
kbringardhey nerdstein16:05
kbringardcome over here16:05
kbringardRazique may be able to help you out16:05
Raziquei'm pm with him ^^16:05
*** po has joined #openstack16:05
kbringardah, OK16:05
nerdsteinkbringard: im good16:05
nerdsteinthanks :)16:05
kbringardthanks… I was trying to help but I'm a little distracted and not much good at the moment :-)16:06
*** obino has quit IRC16:08
*** andreas__ has joined #openstack16:08
Raziquekbringard: it's cool16:10
foexleRazique: verry suspicious issue .... i'll try to find out why the server don't generates fingerprints when i use this keypair16:10
Raziquetake care of urself :)16:10
*** freeflyi1g has joined #openstack16:11
*** ahasenack has quit IRC16:11
*** jedi4ever has joined #openstack16:12
*** kbringard has quit IRC16:13
*** freeflying has quit IRC16:13
_rfzstill can't ping or access my VM's - even tho I have a public ip and internal ip's correctly setup16:18
Raziquedoes glance from 2012.1 use s python 2.7 ?16:18
foexle_rfz: traceroute <ip>16:18
*** bourke has quit IRC16:19
foexle_rfz: look where the package stops16:19
*** bourke has joined #openstack16:19
_rfzfoexle - stops at the VM16:20
foexle_rfz: hmm hmm16:21
foexle_rfz: can you login via ssh ?16:21
*** dwcramer has joined #openstack16:22
_rfzI can on the internal IP16:23
_rfzlet me see if the image has FW rules16:24
*** imsplitbit has joined #openstack16:25
*** dendrobates is now known as dendro-afk16:25
*** rsampaio has joined #openstack16:26
*** dendro-afk is now known as dendrobates16:27
*** kaigan_ has quit IRC16:28
*** jedi4ever has quit IRC16:29
*** adjohn has quit IRC16:31
*** rods has quit IRC16:31
*** Bobbi has joined #openstack16:33
foexlewhere can i find url commands to using swift with keystone ?16:33
*** dobber has quit IRC16:34
BobbiHey guys. I am trying to create an image of an instance with euca-create-image but I get error "Image 1 could not be found". I can confirm that the instance id I am using is valid. The images are also there because I can create new instances using them. Any hints?16:35
_rfzfoexle - i'm losing a lot of packets in the internal network16:37
*** vladimir3p has joined #openstack16:37
foexle_rfz: which network type do you use ?16:38
*** clayg has quit IRC16:40
*** clayg__ is now known as clayg16:40
*** pweiss has joined #openstack16:40
*** clayg_ has joined #openstack16:41
*** dwcramer has quit IRC16:43
*** GheRivero has joined #openstack16:47
foexleRazique: do you know where i can find curl commands to use swift  with keystone ?16:52
foexlei get tokens and all informations but dont know how i can upload files16:52
*** obino has joined #openstack16:53
*** nRy has joined #openstack16:53
*** guigui1 has quit IRC16:54
*** obino has quit IRC16:57
*** obino has joined #openstack16:57
*** krow has quit IRC16:57
Raziquefoexle: I've never used swift16:57
Raziqueshame on me :(16:58
*** krow has joined #openstack16:58
Raziquehave u looked over the docs ?16:58
annegentlefoexle: have you looked at - then go to Troubleshooting chapter and you'll find some curl examles17:01
foexleyes its only with swift client17:01
annegentleexamples that is. They're generic but the token works like X-Auth-Key if memory serves me17:01
*** dendrobates is now known as dendro-afk17:03
*** dolphm has joined #openstack17:03
*** Bobbi has quit IRC17:05
*** ohnoimdead has joined #openstack17:05
*** Eyk^off is now known as Eyk17:06
*** MarkAtwood has joined #openstack17:06
*** negronjl_mobile has joined #openstack17:07
*** negronjl_mobile is now known as negronjl17:07
*** pweiss has quit IRC17:10
*** jog0 has joined #openstack17:11
*** juddm has joined #openstack17:13
*** katkee has quit IRC17:14
*** negronjl_mobile_ has joined #openstack17:14
*** negronjl_mobile_ has quit IRC17:16
*** negronjl has quit IRC17:17
*** jedi4ever has joined #openstack17:18
*** cereal_bars has joined #openstack17:19
*** negronjl_mobile has joined #openstack17:20
*** dolphm has quit IRC17:20
*** rnorwood has joined #openstack17:20
*** negronjl_mobile_ has joined #openstack17:20
*** negronjl_mobile has quit IRC17:21
*** negronjl_mobile_ is now known as negronjl17:21
*** negronjl has joined #openstack17:22
*** Ryan_Lane has joined #openstack17:22
*** oubiwann has joined #openstack17:24
*** jedi4ever has quit IRC17:25
*** negronjl has quit IRC17:25
*** negronjl has joined #openstack17:26
*** rnorwood has quit IRC17:28
*** cereal_bars has quit IRC17:30
*** cereal_b_ has joined #openstack17:30
*** foexle has quit IRC17:31
*** jdg_ has joined #openstack17:34
*** Bobbi has joined #openstack17:35
*** cereal_bars has joined #openstack17:36
*** cereal_b_ has quit IRC17:36
*** jdurgin has joined #openstack17:40
*** adjohn has joined #openstack17:42
*** n0ano has joined #openstack17:45
*** ejat has joined #openstack17:45
*** ejat has joined #openstack17:45
*** n0ano has quit IRC17:45
*** vidd-away is now known as vidd17:45
*** n0ano has joined #openstack17:46
*** MarkAtwood has quit IRC17:47
*** dillon-w has joined #openstack17:52
dillon-whi guys, I really need some help here.17:52
dillon-wwhen i start a instance, it never got allocated ip from network service.17:52
*** maplebed has joined #openstack17:52
*** n8 has joined #openstack17:54
dillon-wfrom the network service log, i can see when network received RPC call from compute, network service then generated another RPC call, which i think isn't needed.17:54
*** n8 is now known as Guest5112017:54
*** dragondm_ has joined #openstack17:54
dillon-wmaybe someone could help me understand the correct work flow for network service?17:55
Guest51120Has anyone seen a potential 'bug' in dashboard whereby the Allocable Storage under System Panel->Services for the node controller does NOT equal the used and physical?17:55
Guest51120dillon-w: I have a log of an instance startup across all services…the only thing it's missing is the glance activities since my instance's image was already cached on machine17:55
*** Guest51120 is now known as n8817:56
n88In my compute node entry in dashboard it reads: Allocable Storage: 9.8TB (18.0GB Used, 144.0GB Physical)17:57
n88… don't quite know how you get 9.8 TB from 144.0GB - 18.0 GB17:58
dillon-wn88 : thx.  i got similar log (that networkservice is working) too.  from that log, i can see when network service received RPCcall to allocate ip, it didn't generate any more RPCcall.17:58
n88dillon-w: here's what I got:
n88dillon-w: I'm guessing you're getting around lines 1.629 and 1.630 (in my time offset) but not the next few lines where IPs are allocated...correct?18:02
*** pweiss has joined #openstack18:02
n88is there anybody here who's working on the dashboard project?18:02
dillon-wn88 : thx.  that's nice log.  how do you generate cross service log?18:03
n88dillon-w: I basically let my cloud sit 'idle' for a while….then fired an instance and grabbed the logs across all the services/machines and then dumped them into Excel18:04
n88I wanted to know all the steps across all services…I also made a flow diagram of the process across the services18:04
uvirtbotNew bug: #850462 in horizon "Pause/Unpause a VM instance" [High,Confirmed]
dillon-wn88 : that's useful, i should do that.  and yes, my network service stops at offset 1.630.18:07
n88dillon-w: so the usual set of questions now is…what type of network mode are you using?18:10
*** darraghb has quit IRC18:10
*** krow1 has joined #openstack18:11
uvirtbotNew bug: #893226 in horizon "double dashboard - dashboard dashboard dashboard" [Undecided,New]
dillon-wn88 vlan18:11
n88dillon-w: ok…I've only been using FlatDHCP so my logs probably won't help you much…sorry should've asked that first…I'm sure some of the steps are the same, but for instance you won't get a floating IP18:12
n88dillon-w: I'm assuming you've got all the right settings in your nova.conf files like vlan_interface18:12
*** egant has quit IRC18:13
dillon-wn88 : yeah, bridge_interface is correctly set.18:13
*** redconnection has quit IRC18:13
viddKiall, you here?18:13
Kiallyea - give me a few mins18:14
*** redconnection has joined #openstack18:14
viddtake whatever time you need =]18:14
uvirtbotNew bug: #893229 in horizon "Allocable Storage Value Incorrect (not matching used, available resources)" [Undecided,New]
*** GheRivero has quit IRC18:16
n88dillon-w: hmm…so you're not able to ping?18:17
dillon-wn88 : ping what?18:17
n88your instances?18:17
n88dillon-w: sorry I came into the chat room mid-convo so I'm not sure what your actual issue is :-)18:18
dillon-wn88 : my instances never got allocated ip.18:18
*** redconnection has quit IRC18:18
dillon-wn88 : network service failed to allocate ip for instance, so compute never actaully start it.18:19
n88dillon-w: yes…if networking doesn't give it an address, it just stops and you get that error msg 'Found 1 instance in db and 0 instances on hypervisor'18:19
n88dillon-w: you created a private network range?18:19
Raziquehey n88 :)18:19
Raziquehey dillon-w :)18:19
dillon-wn88 : rite.18:20
Raziquehey vidd18:20
dillon-whey Razique, really need some help here.18:20
vidd*wave* Razique18:20
n88Here's Kiall's nova.conf template…he uses VLAN:
Raziquealways cool to see u here guys ;D18:20
Raziquedillon-w: let me go home18:20
Raziquei'll be there in 30 mn tops18:20
viddn88 i think i got my scripts all good to go18:20
Raziquedillon-w: is it critical ?18:20
*** Razique has quit IRC18:21
dillon-wRazique : well, i've been debugging for al day. and it's 2 am in the morning. :(18:21
n88dillon-w: you need a fixed_range setting, vlan_intefface, (possibly public interface)…as well as all the correct settings for DHCP18:21
n88(brdige_flagfile, bridge executable, forece_dhcp release if you want it to release internal IPs immediately)18:21
dillon-wn88 : you mean fixed range setting in nova.conf?18:22
n88dillon-w: you do need an internal private network even with VLAN mode18:22
vidddillon-w, in nova.conf and also in nova database18:22
*** adjohn has quit IRC18:22
n88dillon-w: yes you should have a fixed range setting in nova.conf as well as a private network created using the nova-manage command, i.e.: nova-manage entwork create private XX.XX.XX.XX/24 1 25618:23
n88your XX.XX.XX.XX/24 should be similar to your fixed range mask18:23
dillon-wi think i don't have fixed range setting in nova.conf , let me check18:23
*** bengrue has joined #openstack18:24
*** TheOsprey has joined #openstack18:24
uvirtbotNew bug: #893234 in horizon "System Panel Services showing deleted Services" [Undecided,New]
n88the fixed_range setting can be a larger 'set', so you might do —fixed_range= in your nova.conf18:26
n88but your first network you may create as using the nova-manage command18:26
n88( is contained by ''18:27
n88vidd: awesome! Finally got all the pieces working?18:27
*** JesperA has joined #openstack18:27
viddn88, just about =]18:28
*** bcwaldon has quit IRC18:28
*** stuntmachine has quit IRC18:29
dillon-wn88 : well fixed_range covers the network I create18:29
*** jakedahn has joined #openstack18:30
*** bcwaldon has joined #openstack18:31
uvirtbotNew bug: #893236 in horizon ""Error: Bad token, please reauthenticate" going from diablo to essex" [Undecided,New]
*** MarkAtwood has joined #openstack18:31
dillon-wn88 : can you show me how the networks table looks in your setup?18:33
*** Pr0toc0l has joined #openstack18:33
uvirtbotNew bug: #893219 in horizon "clicking logo takes you to login form" [Undecided,New]
dillon-woutput of SQL quiry like 'select * from networks';18:33
*** cereal_bars has quit IRC18:33
*** lorin1 has quit IRC18:35
uvirtbotNew bug: #893243 in nova "Access IP4/6 values ignored in rebuild request" [Undecided,New]
n88the columsn don't line up well in pastebin but you could copy to excel18:36
dillon-wn88 : that's good enough. thx!18:36
*** Ryan_Lane has quit IRC18:39
*** stuntmachine has joined #openstack18:40
*** stuntmachine has quit IRC18:40
*** stuntmachine has joined #openstack18:40
*** Ryan_Lane has joined #openstack18:41
*** scottjg has joined #openstack18:43
*** dolphm has joined #openstack18:46
dillon-wn88 : didn't find any surprise in my network setting compared to yours.18:46
*** CaptTofu has quit IRC18:47
dillon-wn88 : i should dig into source code18:47
n88dillon-w: hmm…and you're using a network switch taht supports vlan tagging?18:47
n88(if not running on a single box?)18:47
*** Ryan_Lane has quit IRC18:47
dillon-wn88 : yeah, it worked before i upgrade to Diablo.18:47
*** Ryan_Lane has joined #openstack18:48
vidddillon-w, did you follow razique's upgrade guide?18:48
dillon-wvidd : i don't know of that.18:48
*** anotherjesse has joined #openstack18:49
*** dachary has quit IRC18:49
vidddillon-w, check that guide and see if you missed something =]18:49
dillon-wvidd : oh, that one.  i looked into that and let me do that again.18:49
*** hezekiah_ has joined #openstack18:49
*** jaypipes has joined #openstack18:50
viddn88 the issue i currently have is that i added the nova-volume stuff after i finished nova and horizon...18:51
viddand now that nova-volume is up (it workes with euca) i cant see it in horizon18:51
viddthe service is listed in the services page, but its not on the menu on the left18:52
n88vidd: you mean you can't see the service in system panel?18:52
n88vidd: crap…wait is volumes supposed to be listed on the left?18:53
n88crap…looks like I'm in the same boat18:53
viddhehe its how you are supposed to assign volumes to instances18:53
n88vidd: but that's odd b/c my nova-volume service was running before i installed horizon18:54
dillon-wvidd n88 do you what '--multi_host' flag does?18:54
*** frankon has quit IRC18:54
n88dillon-w: it allows you to install nova-network on all your nova-compute nodes18:54
*** rods has joined #openstack18:54
n88dillon-w: to prevent a single point of network failure18:54
vidddillon-w, if you have nova-network running on each compute node, you need to set that on18:54
dillon-wi see. so for now, i don't needit.18:54
*** rods has quit IRC18:55
n88dillon-w: are you running nova-compute and nova-volume on a single box?18:55
dillon-wn88 : yea, for debugging purpose18:55
n88vidd: you can still assign volumes via command line...18:55
n88vidd: is that a service you need to have in keystone's services table?18:56
viddn88 yes....but still...its kinda important =]18:56
*** llang629 has joined #openstack18:56
*** llang629 has left #openstack18:56
*** Razique has joined #openstack18:56
n88vidd: it's odd b/c under my System Panel->Services it shows nova-volume enabled and UP18:56
n88and even a way to disable it18:56
dillon-wRazique : hey18:56
Raziquedillon-w: yup18:56
n88vidd: and I guess re-installing dashboard didn't help huh?18:56
viddi never had it there b4 and it worked before...but i dont know if that was from git or if it worked correctly with kiall's ppa =\18:56
Raziqueso what's the status here ?18:56
viddn88 yeah...i have it under the services18:57
dillon-wRazique : so the problem i encounter is, when i start a new instance, 'compute' does a RPC call allocate_for18:57
viddbut there is supposed to be a menu for it as well18:57
Pr0toc0li have a quick question regarding running nova-volume on a separate ubuntu server, what other packages must i install other than nova-volume? nova-common? nova-api?18:58
n88vidd: *sigh* it never ends, does it?18:58
dillon-wallocate_for_instance to network requesting for IP allocation, when network service got that, it doesn't actually do the allocation.18:58
n88vidd: have you looked into dashboard's commands? like reset? purge? clear, etc?18:58
dillon-wbut instead, do another RPC call.18:58
viddn88 and i made the mistake of telling the powers that be what swift can do....i could hear the drool over the phone =]18:58
Raziquedillon-w: ok, can I see both nova-network and nova-compute logs ?18:59
viddn88, im waiting for kiall to become available...i wanted to see if he had the issue before i start playing with things i dont understand =]19:00
RaziquePr0toc0l: nova-volume and iscsitarget is enough19:01
viddKiall, does your dashboard have a menu option for volumes?19:01
Pr0toc0lthank you razique for the clarification19:01
Raziquesure :)19:01
KiallNo. Thats the essex dashboard, and a pre-diablo version19:01
*** dragondm has left #openstack19:02 NO to the playing with stuff i dont understand =]19:02
n88Kiall: Whew! No more trouble then19:02
Kiallvidd, lol19:03
*** stuntmachine has quit IRC19:03
Kialln88, was it you who had the issue upgrading my older nova-volume package after the tgt vs iscsitarget issues?19:04
_rfzI need help :) I can't ping or ssh into my VM's from remote - local IP works fine. I've done opened up ICMP + SSH on nova, the rules are there but nothing19:04
dillon-wRazique : this is for nova-compute19:04
*** dolphm has quit IRC19:04
Raziquedillon-w: weird log :D19:05
*** dpippenger has joined #openstack19:06
dillon-wRazique : and this is for nova-compute
vidd_rfz, you can ping/ssh to the private ip but not the floating ip ?19:06
dillon-wRazique : weird in which way?19:06
Raziquedunno it's like they don't contain enough information19:06
Raziquewhich version are u using ?19:06
_rfzhi vidd, yep that's correct19:06
BobbiHey guys. I am trying to create an image of a running instance with euca-create-image but I get an error: "Image could not be found". I can confirm that the instance id I am using is valid. The referenced image is also available as I can create new instances using it. Any hints?19:08
vidd_rfz, interesting...i just tested mine and im having the same issue =\19:08
*** bcwaldon has quit IRC19:09
viddBobbi, have you tried "snapshot"?19:09
_rfzI've tried with flatdhcp and vlan - both are the same19:09
*** stuntmachine has joined #openstack19:11
Bobbividd, no didn't think of it19:11
n88bobbi: look into the nova manage command19:11
n88nova manage image-create19:11
dillon-wRazique : i'm using diablo RHEL packaging from griddynamic19:12
Raziquedillon-w: ow ok19:12
RaziqueI knew something were slightly different :p19:12
dillon-wRazique : same version works on another setup here.19:12
Bobbin88: thanks, I will give it a try19:13
Raziquedillon-w: can you spawn a new instance and send me both logs while doing it ?19:14
*** adjohn has joined #openstack19:14
_rfzkiall, are you able to ping the public IP address in your setup? (I used your debpackages)19:14
Raziquedillon-w: what is weird here is that the network set up the routes and stuff19:15
dillon-wwell, the log i paste just now is exactly what you wanted19:15
viddn88, i tried to add a security rule via horizon and get "Error adding rule security group: Invalid cidr"19:15
Kiall_rfz, humm>19:15
vidd_rfz, do you have security rules listed for your "default" group?19:15
*** praefect_ has joined #openstack19:16
Kialleverything bar the instance itself can ping the floating IPs, assuming the sec group allows for it19:16
_rfzand same for tcp19:16
Raziquedillon-w: yah sorry i figured19:16
Raziquedillon-w: i'll eat while reading the logs19:17
*** chomping has quit IRC19:17
Raziquebe back soon19:17
dillon-wRazique : thx.19:17
*** javiF has quit IRC19:17
*** krow1 has quit IRC19:17
vidd_rfz, are you using dashboard?19:17
*** cmasseraf has joined #openstack19:17
Kiall_rfz, and where are yout pinging from?19:17
n88vidd: odd…are you admin for that tenant?19:17
*** praefect has quit IRC19:18
viddn88, yes19:18
dillon-wRazique : the weird thing i found is, when network service received the RPC call 'allocate_for_instance' it then do another RPC call.  that's...19:18
*** coli has quit IRC19:18
vidd192.168.0.0/16 worked just fine19:18
n88speaking of admin….does anybody know the set of 'roles' keystone 'supports'? using Kiall's packages I have admin, member, KeystoneAdmin, KeystoneServiceAdmin…are there any more?19:19
vidd_rfz, are you using dashboard?19:19
n88vidd: that's really odd. You sure you didn't mistype the CIDR?19:19
n88have an extra period or slash? space19:19
Kiall_rfz, Keystone supports pretty much any role you want...19:19
_rfzvidd, nah I'm using ecuatools but dashboard works also19:19
Kiallits the other services that need to know what to do with them..19:19
_rfzkiall I'm pinging from my home network19:19
Kialland right now, thats it AFAiK19:19
*** coli has joined #openstack19:19
vidd_rfz, you DO have dashboard turned on though...right?19:20
_rfzvidd - yes19:20
*** JesperA has quit IRC19:20
Kiall_rfz, dashboard makes no difference to ping'ing an instance ;)19:20
Kiallvidd, *19:20
vidd_rfz, check to see if dashboard "inherited" the default security group...mine did not, and i hade the same issue19:20
n88Kiall: Ah…so the roles can be anythign that another openstack 'service' can handle or expects19:20
n88like projectmanager19:21
viddi added it to horizon, issue resolved (dorta)19:21
n88for nova19:21
_rfzthe default security group in dashboard is "default"19:21
vidd_rfz, right...does it have any rules in it19:21
Kialln88, exactly19:21
viddbet you dollars to donutes it does NOT19:21
*** jedi4ever has joined #openstack19:21
n88Kiall: Ok….is there a list of all the roles for each service? Also, is there any way to make new 'roles' by assigning permissions or does that get into code modifications?19:22
*** bonzay is now known as zz_bonzay19:22
_rfzvid it does not19:22
*** ayyansam has joined #openstack19:22
Kialln88, code changes.. and I dont know of such a list19:22
KiallI think it will eventually be easier to make new roles etc.. but I havent looked into the essex keystone updates19:22
*** ayyansam has quit IRC19:23
*** clopez has joined #openstack19:23
n88Kiall: got it. Thanks for the answers19:23
vidd_rfz, i dont care what Kiall says...dashboard having no rules trumps euca (dunno why, it juyst does)19:23
*** jedi4ever has quit IRC19:23
_rfzvidd, I've just added the rules and still nada19:23
Kiallvidd, I can explain that..19:23
*** JesperA has joined #openstack19:23
viddadd your rules via the dashboard and your problem will go away19:23
Kiall2x tenants, euca using 1, dash using another19:23
*** dolphm has joined #openstack19:24
n88Kiall, vidd: yeah there's no way to tell euca what tenant, corrrect?19:24
Kiallboth the euca, nova and dashboard edit the same set of rules. Assuming they are using the same tenant.19:24
n88I've found that the euca-* commands are good for some things and should be avoided at all costs for others :-P19:24
*** lorin1 has joined #openstack19:24
Kialln88, by using the correct credentials, you can19:24
*** lorin1 has quit IRC19:24
*** lorin1 has joined #openstack19:24
n88Kiall: you're right…there is a project name you can set19:24
*** bcwaldon has joined #openstack19:24
Kiallthe euca tools, for admin users, never resrict to a single tenant. if i remember right.. PITA19:25
_rfzVIdd, still same problem19:25
n88Kiall: it's just somewhat annoying too that Openstack components use the concept of 'project' but Dashboard has tenants19:25
vidd_rfz, is that security group assigned to the instance in horizon?19:25
*** sdake has quit IRC19:25
Kialln88, yea, that should be fixed in essex19:25
Kiallremember, that horizon is not an offical project in diablo19:26
*** clauden_ has joined #openstack19:26
*** dirkx_ has joined #openstack19:26
*** Bobbi has quit IRC19:26
_rfzvidd, yep default is assigned to that instance19:27
viddn88, you were right...i messed up the cidr address...had an extra ".0"19:27
vidd_rfz, disassociate the floating ip and then reassign it19:28
viddsee if that helps19:28
*** dendro-afk is now known as dendrobates19:28
*** sdake has joined #openstack19:28
viddKiall, does this mean with essex, we are going to have to do all this all over again?19:29
*** juddm has quit IRC19:29
* vidd is looking to a long July with upgarding to essex19:29
KiallAsk the dev's ;) But - I would be very suprised if it wasnt an easy-ish upgrade19:29
Kiallvidd, book your holiday time ;)19:29
_rfzvidd same thing- still no ping or ssh19:30
*** krow1 has joined #openstack19:30
*** nacx has quit IRC19:31
vidd_rfz, this is going to sound your floating range accesable to your network?19:31
_rfzit is19:31
_rfzI think :)19:31
*** jdg_ has quit IRC19:32
viddfor example, my home network is on the 192.168.15.x range so if my floating ip's are in the 192.168.1.x range, my local network will not be able to access it19:32
*** cp16net has joined #openstack19:32
*** negronjl_mobile has joined #openstack19:33
*** MarkAtwood has quit IRC19:33
_rfzMy floating addies are 200.XX.XX.XX my local network is
vidd_rfz, there is your issue19:33
*** GheRivero has joined #openstack19:33
viddyour loacal network able to see the 200.x.x.x range?19:34
vidd(before i conclude this is actually the issue)19:34
_rfzYeh it should be able to19:35
_rfzi can ping both ranges from withtin the VM19:35
viddwhat is the IP address of the controller node? 200.x.x.x?19:35
*** rnorwood has joined #openstack19:35
*** Ryan_Lane has quit IRC19:36
uvirtbotNew bug: #791444 in horizon "Display keypair in instance detail" [Medium,Confirmed]
vidd_rfz, your VM is INSIDE both ranges19:36
_rfzI've got everything on 1 machine for now19:36
_rfzthe  controller node is 200.xx.xx.xx19:36
viddhrm =\19:37
*** Bobbi has joined #openstack19:37
viddcan a second computer inside your local network ping the controller?19:37
_rfzI reverted back to a single machine setup to elimate errors19:38
_rfzthe 200.xx.xx.xx yep19:38
vidd_rfz, i must say i'm baffled19:39
*** bengrue has quit IRC19:39
*** bengrue has joined #openstack19:40
_rfzwhat type of network setup are you using ?19:40
* vidd is using flatdhcp19:40
_rfzand you have it working?19:41
*** clauden_ has quit IRC19:41
_rfzlet me try flatdhcp19:41
*** sdake has quit IRC19:41
*** dachary has joined #openstack19:41
uvirtbotNew bug: #893270 in horizon "delete  keypair should be "red"" [Undecided,New]
*** clauden_ has joined #openstack19:41
*** sdake has joined #openstack19:41
viddKiall, i still do not understand why horizon wants to use "" for the ip of vnc19:43
*** coli has quit IRC19:43
Kiallbecause thats the defaults for --vncproxy_url=http://78.153.XX.XX:608019:44
*** nelson1234 has quit IRC19:44
*** scottjg has quit IRC19:44
Kiallyou pasted yours yesterday, and it was an invalid URL.. I'm betting is still is ;)19:44
viddbut i changed that19:44
*** nelson1234 has joined #openstack19:44
Kialldouble check it ;)19:44
_rfzvidd/kiall- do you have a nova.conf with a flatdhcp configuration?  I coud use?19:45
praefect_hey, libvirt.xml changed a lot since cactus, now that I use diablo, how can I know I'm using virtio as the network driver for my VMs ?19:46
vidd_rfz, apparently not =]19:46
praefect_the libvirt.xml of the running instance does not list anything except a bridge interface...19:47
dillon-wRazique : it's too late.  i don't want to do a whole nighter.  so I'm leaving for bed... will check with you tomorrow. thank you in advance19:47
*** MarkAtwood has joined #openstack19:50
*** jedi4ever has joined #openstack19:51
*** jedi4ever has quit IRC19:51
*** dillon-w has quit IRC19:52
cmasserafhas anyone ever seen this error?
cmasserafit happens with all nova-manage calls19:52
*** bcwaldon has quit IRC19:53
viddcmasseraf, root should not be making "sudo" calls19:53
viddcmasseraf, what version on nova?19:55
cmasserafwrong copy19:55
swillhas anyone gotten swiftauth to work with swift beyond just the admin account?  I have the admin account working, but no other account works (500 Error).19:55
cmasserafbut the newest in the apt-get19:56
*** mdomsch has quit IRC19:56
*** Ryan_Lane has joined #openstack19:56
viddcmasseraf, did you add Kiall 's ppa which has fixes for most know issues?19:57
swillI have confirmed that using the swift -A command will create tokens on the keystone side.19:57
swilli can then do curl commands to get auth details for the token it creates, but the swift command returns an error.19:57
*** hggdh has quit IRC19:58
cmasserafwhat do I add?19:58
viddcmasseraf, are you also trying to integrate keystone?19:58
viddcmasseraf, check out
viddcmasseraf, do you plan to include keystone?20:00
*** mdomsch has joined #openstack20:01
phschwartzOk, I am back to square one. I can launch instances, but they all fail to get cloud info from The nat rule is in iptables, but I am not sure why it is doing it. Anyone able to point me in the right direction?20:01
cmasserafbut I'm trying to get the openstack to run before adding anything (I'm trying withou success since last week)20:02
viddcmasseraf, then i highly recommed following Kiall 's scrpts to get it all running20:02
viddwith good connection speeds, all done in < 3hours20:03
cmasserafok, the thing is... it uses 2 NIC and I only have one in the servers (it's a university lab cluster for computations)20:04
phschwartzcmasseraf: I ran into the same issue. I just set both to the same nic and I can spawn instances20:05
*** hggdh has joined #openstack20:05
viddmy scripts are highly experimental, and are designed for single eth20:05
viddbut are based off Kiall 's scripts20:05
phschwartzI am not sure if the issue I am having with the access to is related to the fact that I have one nic.20:06
*** jedi4ever has joined #openstack20:06
cmasserafphschwartz: oh, perfect... then they all get assigned statically or by DHCP?20:06
*** jedi4ever has quit IRC20:06
*** redconnection has joined #openstack20:06
phschwartzActually, it is doing it via vlan with a fixed ip range.20:06
phschwartzWorks nicely for it at the moment.20:06
viddfor one eth per server i recommend flat networking...not vlan20:06
*** Bobbi has quit IRC20:07
*** dnjaramba has quit IRC20:08
viddKiall, so i set --vncproxy_url=http://$nova_host:6080 and now the vnc console does not come up at all20:09
viddshould i take down the --vncproxy_port=608020:09
KiallWell, I dont have that setting.. so maybe?20:10
vidd$nova-host is the same machine running NoVNC20:10
KiallI think thats the default BTW20:10
viddand you said --vncproxy_host= ...right?20:10
*** lorin11 has joined #openstack20:10
Kiallall i have in my config re VNC is: --vncserver_host=
Kiall --vncproxy_url=http://%HOST_IP%:608020:11
viddsweet =]20:12
viddwe have contact =]20:12
*** lorin1 has quit IRC20:12
Kiallwhat was it?20:12
*** mdomsch has quit IRC20:13
viddi dropped the --vncproxy_port20:13
Kiallweird.. wonder why that caused it20:13
viddand added the --vncproxy_url20:14
Kialloh, vncproxy_url wasnt set? that explains it ;)20:14
*** rnorwood has quit IRC20:14
viddmy thinking is it was doing vncproxy_url + vncproxy_port20:15
*** rnorwood has joined #openstack20:15
viddso making an address like 192.168.X.X:6080:608020:15
*** rnorwood has quit IRC20:15
Kiallhumm. weird..20:15
Kiallrmk, ping20:16
*** dotdevops has joined #openstack20:16
viddwhen i had no vncproxy_url, but did have vncproxy_port, i was getting url with
viddso i had to replace the ip with the correct IP20:16
viddwhen i added the vncproxy_url and left the vncproxy_port, i was getting a timeouit20:17
KiallOkay - so vncproxy_port is the port the service binds on20:17
Kiallwhile vncproxy_url is the full URL (inc :6080) to access the service20:18
viddright...but apearently you cannot set the URL AND the port20:18
phschwartzSo I found my issue. for single nic with kiall's in vlan mode, you can't use the localhost ip, you need the public ip of the system. so no
phschwartzinstance came up perfect with it set to the real ip.20:18
Kiallvidd, actually, both are needed, vncproxy_port defaults to 6080 while vncproxy_url defaults to
viddphschwartz, did you look at my scripts at all?20:19
Kiallphschwartz, dooh.. yea.. avoid 127.0.0. pretty much everywhere ;)20:19
viddKiall, that is obviously NOT correct...i had both and it timed out20:20
phschwartzyes, Issue I see is the actual configuration differences between single nic with vlan and flat is not far off. I modified with 2 config flags from vlan to flat after the install.20:20
Kiallvidd, they are used by 2 different parts of nova20:20
* Kiall adds --vncproxy_port=6080 to test20:20
viddKiall, you said 10 minutes ago that you do NOT have --vncproxy_port set in your .conf20:21
KiallI dont..20:21
Kiallhence why I'm adding it to test20:21
KiallBut - I've read the code, and I belive both are needed20:21
KiallIt just so happens the default --vncproxy_port was the correct value for me20:21
phschwartzto have vnc to it, is there any special services I need to install? your talk has me intrigued :)20:21
* vidd read the code too [have no idea how it works....but ive read it ]20:21
Kiallphschwartz, your using my packages you said?20:22
phschwartzkiall: your latest correct20:22
KiallIf you are.. just openstack-novnc and nova-vncproxy on the controller node20:22
phschwartzused the all in one script.20:22
Kiall(or the VNC node if you have lots of hardware)20:22
Kiallphschwartz, then VNC should already be working20:22
*** adjohn has quit IRC20:23
*** anotherjesse_ has joined #openstack20:23
*** adjohn has joined #openstack20:23
viddphschwartz, im forking Kiall 's code for my own uses (like having different database usernames and passwords)20:23
*** dirkx_ has quit IRC20:23
n88Kiall: So then I just point my web browser to the VNC URL using say chrome?20:23
phschwartzOnly issue I see is your fork isn't in sync with his. He has made some nice changes to his recently that looked to be missing from yours.20:24
* vidd does not like the "use root for your database usernames and let root be available from everywhere20:24
* Kiall doesnt mind it for "install quick", them customize to suit20:24
phschwartzkiall: hmm openstack-novnc is installed, nova-vncproxy is not running and stops the moment it is started.20:25
*** anotherjesse has quit IRC20:25
*** anotherjesse_ is now known as anotherjesse20:25
Kiallphschwartz, any error?20:25
viddphschwartz, what changes do you see that i dont have?20:25
* vidd has been busy ironing out bugs to add new stuff20:25
phschwartzKiall: not that I see in the nova-vnc log20:26
Kiallright soo.. dpkg -l | grep nova-vncproxy ?20:26
phschwartzii  nova-vncproxy                    2011.3+20111111-0mit3                   OpenStack Compute - VNC proxy20:27
Kiallthats the latest package alright.. weird20:27
n88Kiall: how do I get the URL from the API…the vnc doc page doesn't have the right command20:27
Kialln88, I've never actually needed it from the API.. Always just clicked the dashboard link ;)20:28
phschwartzmeh, I will leave it at this. I never have the need for a graphical install so no need for me to worry about it.20:28
*** nelson1234 has quit IRC20:28
Kiallvidd, well.. for a moment, I thought you were right20:28
KiallBut instead you've shown up a bug in my firewall rules ;)20:29
KiallNeed to allow 6080 without the VPN connection ;)20:29
viddhehe then perhaps i have the same bug20:29
*** dirkx_ has joined #openstack20:29
n88Kiall: oh…gotcha. The 'in progress' doc page mentions a bogus command that you should be able to use from the nova api to get the url20:30
n88but I will look at it via dashboard20:30
*** JesperA has quit IRC20:31
KiallYea - Cant get VNC going from outside the office ;) Not sure what port I've missed..20:31
*** negronjl_mobile has quit IRC20:31
KiallOMG. Thats a test instance on a node a shutdown earlier -_-20:32
KiallNo wonder20:32
*** nelson1234 has joined #openstack20:33
Kiallvidd, yup. it 100% works with both set20:33
*** dendrobates is now known as dendro-afk20:34
*** JesperA has joined #openstack20:35
viddwhere did you set the firewall rule?20:35
Kiallon our firewalls ;)20:36
viddso NOT the controller20:36
Kiallwe have a pair of cisco FW's front and center, it was blocking 6080 from the outside world20:37
*** sfvg has joined #openstack20:37
*** pweiss has quit IRC20:38
*** GheRivero has quit IRC20:38
*** dprince has quit IRC20:39
*** n0ano has quit IRC20:39
sfvghello....what is the best script for testing openstack in a vm?20:39
*** dendro-afk is now known as dendrobates20:40
Kiallsfvg, devstack is great + well tested for a quick and dirty test20:40
viddKiall, i dont understand why it works for you and not for me =\20:40
*** cereal_bars has joined #openstack20:40
Kialladd it back, and i'm betting it will still work ;)20:40
Kiallyou probably had a typo or something!20:40
KiallWouldnt be first time20:40
KiallI seem to remember a "--vncproxy_url=" ;)20:41
sfvgkiall: thanks...once I get past devstack...what is the best scripts for getting everything installed for a proof of concept?20:41
viddKiall, there was never a --vncproxy_url= it just wasnt there20:41
Kiallsfvg, that depends I suppose, if you want keystone+the dashboard, probably mine ;) Othewise, The stock ubuntu packages work well20:42
sfvgkiall: i need to test the entire stack20:42
swillwow this is messed up.  in swift_auth, i have narrowed down my error to the line conn = http_connect(...).  the strange thing is that if i run the same exact thing from a python console it works, but it fails in the script.20:42
Kiallwell, this gives you keystone+dashboard+nova+glance on ubuntu oeniric
KiallI've never had a need for swift, hence there is nothing in that repo for it20:44
sfvgkiall: this diablo?20:44
Kiallyea - its the stable/diablo branch20:44
*** GheRivero has joined #openstack20:44
sfvgperfect...i do need to test swift also...can I just add that?20:44
KiallYea - It should be possible to just set that up manually, its not required by anything else so it should be find to add after the fact20:45
viddKiall, yes...they work together...horizon jut thought compute and nova-network was down -=20:45
viddrestarted both, all is well and good in the world20:45
sfvgkiall...thanks for all the info20:46
KiallIf you want a quick and dirty OS test with that, give ./ a go in a VM with 2x NIC's eth0 = something public, and eth1 just needs to exist for a single server test with my scripts20:46
Kiall(edit the settings file first...)20:46
viddKiall, i think mys cripts are ready for prime-time =]20:46
sfvgok..i will give that a go20:46
Kiallvidd, we'll see ;)20:46
viddKiall, going to give them a go?20:47
KiallYea - If they actually work yet ;)20:47
viddjust pushed the vnc-proxy fix =]20:48
viddshould be good to go20:48
*** odyi has quit IRC20:48
viddi have not added all your new-and-improved glance uploads yet though20:49
*** krow1 has quit IRC20:49
n88vidd/Kiall: for vnc this document is not entirely correct, but does highlight the settings/flags correctly I think:
viddand these scripts are currently designed for single eth20:49
Kialln88, I've had VNC working for ages ;) Was just checking something for vidd20:50
* vidd needs to go snooping through Kiall 's scripts for more improvements =]20:52
KiallI dont think I've chnaged anything other than the default images20:53
viddyou fixed your broken ec2 creds20:53
rmkKiall: here now20:53
Kialldid i?20:54
Kiallrmk, where was that PPA of your's buried with the replacement vncproxy?20:54
viddkeystone-manage credentials add admin EC2 'admin:admin' %ADMIN_PASSWORD% admin is not going to work i think20:54
viddthe ' marks will be put into the database20:54
Kiallvidd, no they wont...20:55
viddshould probably be keystone-manage credentials add admin EC2 admin:admin %ADMIN_PASSWORD% admin20:55
rmkKiall: It's an ugly package -- you're welcome to redo any part of it20:55
Kiallmeh .. the quotes are fine.. standard bash parameter handling20:55
*** krow1 has joined #openstack20:56
rmkKiall: btw your glance build doesn't seem to auth against keystone without passing a token via -A20:56
Kiallrmk, I just wanted to have a quick go and see how it works in chrome20:56
rmkKiall: cool -- yeah it should "just work"20:56
Kiallrmk, I was under the impression that was required?20:56
rmkKiall: Nope, should not be required if the OS_ variables are present20:56
rmkThere was a fix for that somewhere, I thought it made its way into stable.20:57
viddrmk, i dont hink Kiall passes the OS_ variables20:57
Kiallwell, its the latest version of stable/diablo .. at least, it was a few days ago20:57
Kiallvidd, I do in my novarc file.. Just not in the scripted stuff20:57
KiallAnyway - rmk, you wouldnt remember fix? I'm pretty sure there has been no stable/diablo fixes for glance until very recently..20:58
KiallAnd thats if any of them have made it past gerrit yet..]20:58
rmkKiall: Your last glance build was 10/2120:59
rmkKiall: There are a bunch of commmits there now20:59
Kiallrmk, they were only merged very recently :)20:59
rmkyeah just noticed21:00
rmkchecking if any of those commits fix it21:00
KiallBut since they are cherry picked from master, they look older21:00
Kiallrmk, it looks like they are *very* recent..21:01
Kiallas in the oldest chance was merged Mon, 21 Nov 2011 18:55:26 +000021:02
Kiallthats like.. 3 hours ago ;)]21:02
*** llang629_ has joined #openstack21:02
rmkHrm, look at the nov18 commits.21:02
rmkThere are patches from a while ago.21:02
rmkCommitted recently, authored a month ago.21:03
KiallThey were comitted to master a while ago21:03
Kiallmerged to stable/diablo a few hours ago ;)]21:03
rmkah right21:03
*** mikeyp has joined #openstack21:03
*** PotHix has quit IRC21:04
*** mikeyp has quit IRC21:06
rmkThis might just be a pipeline issue21:06
*** MarkAtwood has quit IRC21:07
Kiallrmk, really? Let me know what you find!21:07
*** mikeyp has joined #openstack21:07
*** mikeyp has left #openstack21:07
*** nerdstein has quit IRC21:09
*** hggdh has quit IRC21:10
*** jj0hns0n has joined #openstack21:10
*** Razique has quit IRC21:11
*** hggdh has joined #openstack21:16
*** llang629_ has left #openstack21:16
*** krow1 has quit IRC21:19
*** clopez has left #openstack21:19
*** mikeyp has joined #openstack21:20
*** krow1 has joined #openstack21:22
*** martines has quit IRC21:24
*** mgoldmann has quit IRC21:25
*** andrewbogott has quit IRC21:25
*** primeministerp has quit IRC21:28
uvirtbotNew bug: #893333 in openstack-integration-tests "Add test get large object in " [Undecided,New]
*** andrewbogott has joined #openstack21:31
*** martines has joined #openstack21:32
*** lorin11 has quit IRC21:37
*** miclorb_ has joined #openstack21:37
*** MarkAtwood has joined #openstack21:38
rmkIs anyone using cloudpipe?21:38
_rfzVidd - you are 100% right - you need to add the rules to the seurity groups via dashboard21:39
_rfzI've got flatdhcp working21:40
_rfzrmk not yet21:41
Kiall_rfz, no no no. vidd is absolutely not correct! The dashboard+euca2ools+novaclient all edit the same set of rules.21:42
KiallAssuming you are logged into the same tenant21:42
_rfzlet me see21:42
_rfzI'm tennant 121:42
_rfzon dashboard21:42
*** praefect_ has quit IRC21:43
Kiall_rfz, were you using the euca tools? or novalcient?21:43
_rfzI've your setup21:43
*** hezekiah_ has left #openstack21:43
_rfzI can't type21:43
_rfzI've used your packages kiall21:43
_rfzso all the end points are as per you've defined21:44
_rfzand I when I use ecua2ools they rules don't do anything21:44
*** dendrobates is now known as dendro-afk21:44
KiallDo you only have 1 tenant?21:44
*** clauden_ has quit IRC21:45
*** clauden_ has joined #openstack21:45
_rfzadmin, demo and one of mine21:45
_rfzall tenants and endpoints have been created from your script21:45
Kiallthen euca is connected to one of those other tenants21:46
Kiallgo check their default groups for those rules you added ;)21:46
_rfztennant 121:46
*** clauden_ has left #openstack21:46
_rfzthat is with euca2ools21:46
_rfzvia dashboard it's 221:46
_rfzso it's demo21:47
Kiallyea - so euca2ools was connected to the wrong tenant, assuming you wanted them to talk to the "demo" rather than "admin" tenant21:47
_rfzYes, I see your right21:47
_rfz100% right21:48
_rfzI was logged in as demo21:48
Kiallvidd was saying that adding rules via the CLI doesnt work.. It does.. Just euca2ools doesnt have a tenant dropdown, you need the right env variables21:48
_rfzI've just logged in as admin (1) and I can see the rules I created via euca2ools21:48
_rfzwhat a head f**k :)21:49
KiallIf you switch your EC2* variables to something associated with demo (2), they will appear there instead...21:49
vidd_rfz, what you add in euca will ONLY affect the specific user you have in your env at the time you do it21:49
Kiallexactly - thats not what you said earlier though ;)21:50
_rfzYep I understand now21:50
_rfzand my euca2ools env is tennant 121:50
viddKiall, am i now no longer allowed to learn? =]21:50
_rfzlet me try bring up another node21:50
Kiallvidd, nope. from now on, you get beaten each time you slip up!21:51
* vidd has gained a greater understanding [and hatred] of horizon21:51
Kiall_rfz, you in? ;)21:51
Kiallvidd, lol .. Overcomplicating stuff again? ;)21:51
_rfzI'm in for the learning :)21:52
viddKiall, it IS complicated21:52
viddmethinks you oversimplify21:52
Kiallhorizon has exactly 3 setting that ever needs changing DB, Swift=off, Quantum=Off, how is that complicated? ;)21:52
_rfzDo I need to create a bridge device on my compute nodes also?21:53
Kiall_rfz, I thought you were using VLAN mode?21:53
_rfzI gave up :)21:53
Kiallah.. fair enough21:53
_rfzand switched to FLAT21:53
Kiallthen yea, every node needs it21:53
rmkHorizon can't really be blamed for much.21:53
viddKiall, its not horizon...its keystone21:53
viddkeystone is evil21:54
*** jdg has joined #openstack21:54
rmkI have one major problem with Horizon and that's when you have a lot of instances.21:54
_rfzKiall, I don't think my switch has vlan tagging activated - so it won't work anyway21:54
rmkIt just gets progressively slower.21:54
jdgHas anybody tried using the script on a physical machine rather than a VM?21:54
WormManthat's ok, the euca API service gets really angry when you have lots of instances :)21:55
viddjdg, yes...i a month or two ago21:55
jdgvidd:  How did it go?  Recommend, or run away?21:55
*** stuntmachine has quit IRC21:55
viddjdg what doi you plan to do after it runs?21:55
jdgI've been having a heck of a time getting everything configured/working manually.21:56
viddif you are going for the learning experience...go for it21:56
jdgWhat I was thinking was I'd set it up then copy the nova.conf to a second machine and add it as a compute node21:56
jdgI've used it in VM's a number of times with good succes.21:56
*** MarkAtwood has quit IRC21:56
jdgMy physical installs however have been another story.21:56
viddif you want it to do actual work, go with kiall's script (or mine if your adventurous)21:56
jdgvidd:   Hmmm... which script is Kialls?21:57
*** rnorwood has joined #openstack21:57
Kialljdg, devstack is short for development stack .. Its completelty and utterly unsuitable for production use.. even as a base for production use..21:57
*** dgags has quit IRC21:57
vidd jdg21:57
KiallI put my install scripts up here ..
*** MarkAtwood has joined #openstack21:57
Kiallthey install eveything from packages, rather than git clones etc...21:58
_rfzjdg - use Kialls repo - it works great21:58
viddmy scripts use Kiall 's repo's as well21:58
*** miclorb__ has joined #openstack21:58
_rfzand vidd's work great also :)21:59
Kiall_rfz, actually.. you just setup FlatDHCP with my scripts? or just with my repo?21:59
jdgvidd:  Can you point me to said script?21:59
jdgOr Kiall... didn't see you there.  :)21:59
_rfzI used your repo to setup openstack with VLAN, it didn't work so I did the nova.conf manually, reading the network documentation and reading vidd's scripts21:59
KiallSure, All my stuff is @ ..22:00
Kiall_rfz, ah okay22:00
_rfzOnce I have FLAT working I will go back to vlan22:00
KiallNever gonna happen ;)22:00
*** miclorb_ has quit IRC22:00
jdgOk, thanks all.  I will take a look at the two of these and maybe start over with one of them.22:00
viddKiall, my scripts do work22:00
jdgHopefully things will go better this time!22:00
Kiallvidd, huh?22:01
Kiallvidd, oh yea.. I was joking earlier ;)22:01
*** imsplitbit has quit IRC22:01
viddi know =]22:01
rmkCloudpipe documentation references "wget -O /tmp/payload.b64", to run out of a script in the VM.  Question is, what is that IP?  Looks like a link local address but I don't see anything of the sort up on my hypervisor nodes.22:02
viddjdg, if you have any issues ...or comments...or recomendations regarding my scripts (if you choose to try them) let me know22:02
Kiallrmk, its the EC2 metadata URL22:02
Kialleach nova-network node NAT's that to the EC2 api on the nova-api server22:03
jdgvidd:  So maybe I should ask beforehand...22:03
jdgI'm attempting to set up a Cluster Controller with multiple compute nodes.22:03
* vidd will soon be investigating adding a vlan option to it22:03
jdgI have 2 Nics in each system, one 1GB and one 10GB22:03
jdgThe idea was that the 10GB would be use for iSCSI volumes22:03
uvirtbotNew bug: #843049 in keystone "Verify: feedback/fixes from Bryan addressed" [High,Confirmed]
rmkKiall: ok I actually do see it, I need to stop using ifconfig ... ever.  It shows up via ip addr.22:04
jdgDoes that sound like something your script might be able to help me with?22:04
viddwhich eth port is which?22:05
jdgeht0 is 1Gig and eth2 is 10Gig22:05
jdgeht0 == eth022:05
viddno eth1?22:05
Kiallvidd, lol.. tell me you didnt hardcode that in your scripts? lol ;)22:06
Kiallb rb22:06
jdgIt's present but no cable22:06
rmkKiall: So, is it doing pat also to port 80 on the api server?22:06
_rfzAlright! 2 nodes working :)22:06
jdgProblem is I'm out of ports on my 1 GB switch :)22:06
viddKiall, yes it is hard-coded22:06
rmkKiall: In other words, 80 on link local to ec2 port on the api server?22:07
*** nerens has quit IRC22:07
*** anotherjesse has quit IRC22:07
viddjdg, you will need to modify the nova-settings file22:07
jdgOk, just swap eth1 to eth2 correct?22:08
viddrun my to build the nova-settings file, then modify it to use vlan and to set the eth ports properly22:08
jdgOk, I'm going to give it a go this afternoon.  One other question, I don't see any requirements regarding Ubuntu version?  11.10 Ok or should I use something else?22:09
viddi adjusted the scripts to look to see if there is a nova-settings file first, and give you the option to use it, modify it or destroy it and use my defaults =]22:09
*** GheRivero has quit IRC22:10
viddi went with the latest22:10
viddit has NOT been tested on 11.0422:10
_rfzhmm - I've added the iptables rule to the compute node, but i'm still getting this error[WARNING]: '' failed: url error [[Errno 111]22:10
jdgvidd:  Ok, so 11.10 it is.  I'll get the system imaged and get going.  I'll let ya know how it goes.  Thanks!!22:11
Kiallrmk, 80 on 169.254 -> $nova_ip:877322:11
viddill be here for at least another 5 hours so if you run into any questions ill be here =]22:11
*** deshantm_laptop_ has joined #openstack22:11
rmkKiall: Nods testing it now.  Thanks, that really did clear it up for me.22:12
Kiall_rfz, is it seperate compute node and controller node?22:12
_rfzkiall - yes22:12
Kialldoes each one take a few seconds to timeout? or does it whizz through the 100 or so attempts?22:13
_rfz100 or so attempts22:13
viddjdg, also, make sure you run the "" before you run the nova-preload script =]22:13
Kiall_rfz, but - do each attempt take about 2 seconds? or are the instant-ish?22:14
_rfzkiall - sorry it takes a few seconds on each to time out22:14
KiallOkay.. Weird, I personally install nova-compute, nova-network and nova-api on each of the nodes22:15
*** deshantm_laptop has quit IRC22:15
Kiallbut, If the NAT rule is in place, it should work...22:15
*** lts has quit IRC22:16
rmk nova meta 141 set foo=bar -- complains that the server doesn't exist, which it of course does22:17
rmkKiall: Can you see if you can set metadata?22:17
Kiallrmk, its a bug with stable/diablo as far as I know..22:17
KiallI've not had a need for it so havent looked too far into it.. but it gave the same error for me22:18
*** Ryan_Lane1 has joined #openstack22:18
*** Ryan_Lane has quit IRC22:18
*** Ryan_Lane1 has quit IRC22:18
*** Ryan_Lane1 has joined #openstack22:18
*** Ryan_Lane1 is now known as Ryan_Lane22:18
viddKiall, what is the benifit of vlan over flat?22:18
viddbesides support for more than 1 network22:18
Kiallsupport for more than 1 network isn't enough?22:19
Glaceevidd: I would say.. customer segmentation?22:19
Glaceesecurity.. if you dont use sec groups22:19
rmkKiall: Don't suppose you know of any patches?  I'll go digging if you don't know off hand.22:20
KiallI havent looked to be honest22:20
Kialli tried it out of curisoity, but havent had an actual need so moved on22:20
viddfor my current use case, im creating a private cloud for a single user22:20
rmkKiall: I can see a lot of use cases for this -- but mainly Cloudpipe.22:20
_rfzkiall - i don't have novi-api installed on my compute node22:20
rmkMaybe I can add it via nova-manage.22:20
viddexternal customers will be on a different cloud controller and nodes which will of couse need to be vlan22:21
*** dolphm has quit IRC22:21
Kiallsee - planning on running 2x openstacks sounds silly to me, unless of course you are doing geo redundancy...22:21
viddbut if only one customer will have access to all services flat is fine22:21
*** sfvg has quit IRC22:22
Glaceevidd: If you dont need specific cidrs.. I would go flat22:22
viddKiall, the 2x openstacks are "company A" and "company B"22:23
*** haji has joined #openstack22:23
viddfinacial isolation =]22:24
Kiallvidd, sounds like both will waste money on the overheads ;)22:24
*** GheRivero has joined #openstack22:25
viddmy company will need to have the ability part ways if the partnership they are involved in goes south =]22:25
hajihey kiall, i see you updated the scripts, any important changes?22:26
Kiallvidd, sounds like your company is doing a pile of work for something that will never work ;)22:26
*** GheRivero has quit IRC22:26
viddand having "thier stuff" and "our stuff" running on the same OS stack has the potential for issues22:26
Kiallhaji, the only real changes were the image uploads..22:26
*** clauden has joined #openstack22:26
KiallThe older way used the .tar.gz ubuntu downloads, which aren't 100% suitable for openstack.22:27
Kiall(You can't upgrade the kernel inside them.. Whatebver version is has is what its going to stay!)22:28
KiallThe newer one's let you "apt-get upgrade" the kernel inside the instance correctly22:28
stevegjacobsKiall, whats the link to your scripts again?22:28
KiallIts safe to delete the current images you have (it wont affect running VMs based on them), and re-run ./ && ./ && ./glance-upload-lucid-loader.sh22:29
KiallAnyway - Back to my TV.. IRC keeps popping up on-top ;)22:30
*** jdg has quit IRC22:30
viddKiall, quick question....will your pin servive a do-release-upgrade?22:31
Kialldo-release-upgrade is to go between 2 ubuntu versions?22:32
KiallNo idea what that has to do with my packages!22:32
*** CaptTofu has joined #openstack22:32
viddwhen you do a do-release-upgrade, all packages get updated to the newer virsion in that release22:33
KiallOh - will it survive, I've no clue. Frankly the last time i trusted an "upgrade" was about 8 years ago.22:33
viddso will your pin keep us with your ppa or will new ubuntu ones over-ride them22:33
KiallI've quite literally not "upgraded" a server in 8 years. I reinstall them ;)22:33
KiallIf the pin package is kept, then yes, if do-release-upgrade removes it.. then no22:33
KiallYou want to upgrade to precise already??22:34 looking to 12.0422:34
*** markvoelker has quit IRC22:34
viddyou know how much confidence i have in the *buntu boys to get this right =]22:35
KiallOh, when 12.04 rolls around, I'm hoping my packages won't be needed. If they are, I'll update the PPA to include precise packages22:35
cmasserafafter installing openstack throught Kiall script what would be the default dashboard user?22:35
KiallI would imaging they should survive the upgrade22:35
Kiallcmasseraf, "admin"22:35
cmasserafwith pass = to the one setted in the settings file right?22:36
*** sdake has quit IRC22:39 next task it to figure out the best way to take a running-on-bare-metal server and "convert" it to a vm image to run on OS22:40
viddwithout destroying the original =]22:41
Kiallboot from CD22:41
Kiallvidd, dd if=/dev/sda1 of=/usbdrive/some_file.img22:41
Kiallqemu-img conver -c -O qcow2 /usbdrive/some_file.img /usbdrive/some_file.qcow222:42
Kiallqemu-img convert -c -O qcow2 /usbdrive/some_file.img /usbdrive/some_file.qcow2*22:42
Kiallthat should handle the very very basics for a linux server...22:42
*** rsampaio has quit IRC22:42
*** rnorwood1 has joined #openstack22:43
*** sdake has joined #openstack22:43
KiallBut really - You're doing it wrong ;) Re-provision the server, use this as a test of your deployment docs etc etc22:43
*** GheRivero has joined #openstack22:43
Kiall(or, more likely, use it to write those deployment docs)22:43
*** rnorwood has quit IRC22:43
viddKiall, im not sure im following you22:44
Kiallcreate a raw disk image of the / partition, then convert from that to a compressed qcow2 image to shrink it22:45
viddtake server down first22:45
Kiallthen edit all sorts of crap like fstab, network stuff etc etc22:45
viddboot from a live cd22:45
Kiallthen upload to glance exactly the same way as my oneiric script22:45
viddthen run the if=/of=22:46
viddupload to glance/swift22:46
Kiallwell - edit some stuff in the image somewhere between if=/of= and the upload...22:47
* vidd has thrown that part into the "convert" heading =]22:47
viddbut this SHOULD leave the original server intact, and preserve hashed passwords etc22:48
Kiallbasically, make sure /etc/udev/rules.d/* is deleted, ensure it uses DHCP, and a script to download SSH keys from 169.254 couldn't hurt... (or install cloud init_22:48
viddin the new virtual image22:48
viddnot all these are running ubuntu22:48
KiallYea - The original server will "think" it just had a reboot, and the image of the server should boot up, get a DHCP address and sing along as if nothing happened.22:48
viddmost are etch [or older] debian22:48
viddim not too sure i want to even play with the windows servers22:49
*** negronjl has quit IRC22:49
KiallThe hard part is getting the image to boot and get network, which is basically just "rm /etc/udev/rules.d/70-pers*", giving the partition a label (eg e2label /dev/loop0 cloudimg-root) and editing thr Fftab to detected the root via LABEL=cloudimg-root22:50
KiallThe windows server.. are umm... going to be harder ;)22:50
KiallAnyway - You should probably reprovision the servers, rather than trying to clone them...22:51
_rfzdo you need to install nova-api on a compute node?22:51
viddKiall, what i "should" do and what they "want" me to do are not always the same thing =]22:52
*** krow1 has quit IRC22:52
viddbut there are user accounts on some of these machines we have no records of that we inherited from other companies we urchased in the past22:53
viddso reprovisioning will be "problematic" at the least22:53
Kialllol.. nice22:54
stevegjacobsKiall, I want to update my lucid image - there are several new upload scripts in your setup22:54
stevegjacobswhich do I use?22:54
Kiallstevegjacobs, yea.. i should document them :)22:54
Kiall./ followed by ./glance-upload-lucid-loader.sh22:54
viddlast time we had to do a forced reprovisioning [hardware failure] the phone would stop ringing for a month22:55
Kiallthe loader can be reused for other pre-11.04 images, hence its kinda seperate22:55
*** bsza has joined #openstack22:55
*** GheRivero has quit IRC22:55
cmasserafwhats the equivalent to euca-describe-instances in keystone?22:55
*** dolphm has joined #openstack22:56
viddcmasseraf, keystone does not do that kind of work22:56
*** chadh has quit IRC22:56
*** Shentonfreude has quit IRC22:56
viddkeystone is more like the traffic cop that say "you can go" and "you cant go"22:57
cmasserafoh... i see22:58
cmasserafI misunderstood it's working...22:58
*** syah has quit IRC22:58
viddcmasseraf, what you need to do is get your EC2 creds into keystone so that when euca checks with keystone, they get the green light to run the command22:58
cmasserafeuca is not installed22:58
*** rnorwood1 is now known as rnorwood22:59
cmasserafI just finished installing and the image halts on Build state22:59
cmasseraf(started thtough the dashboard)23:00
*** bsza has quit IRC23:00
*** Eyk is now known as Eyk^off23:01
viddcmasseraf, try browsing away and comoinmg back23:02
cmasseraftryed, it's been like this for 15 minutes already23:03
*** chadh has joined #openstack23:03
*** negronjl has joined #openstack23:03
viddif you dont have euca installed, then try nova client23:04
viddKiall, 's script install that but i dont know the command for it23:04
*** negronjl has quit IRC23:05
viddcmasseraf, you did use Kiall 's scripts right?23:05
*** negronjl has joined #openstack23:05
viddi know you didnt use mine (otherwise euca would be installed)23:05
stevegjacobscmasseraf - the command to see the status of all instances is 'nova list'23:06
cmasserafI used kiall's23:06
stevegjacobsthen 'nova show <ID>' to see detailed info23:06
cmasserafit's not launching23:12
cmasserafit keeps pending23:13
Kiallcmasseraf, check the nova-compute and nova-network log files23:15
Kiallthey will likely tell you whats wrong (even if its a tad cryptic sometimes ;)23:16
_rfzOky, I got the passed the but it only works when the VM has a public ip assigned to it23:18
*** chadh has quit IRC23:18
Kiall_rfz, aha.. i know the issue, 1 sec23:19
Kiallgrab the DMZ and Fixed range setting from this:
Kialland set them to your fixed IP range23:20
Kiallrestart everything, and it should work23:20
_rfzokay let me try :23:20
cmasserafKiall: there's a exception in nova-compute:
*** PeteDaGuru has quit IRC23:25
*** TheOsprey has quit IRC23:28
*** llang629_ has joined #openstack23:29
_rfzkiall - so I put my fixed ranged in dmz_net ( and netmask in dmz_mask ( ?23:29
*** vidd-away has joined #openstack23:31
*** vidd-away has joined #openstack23:31
*** rustam has quit IRC23:31
*** vidd has quit IRC23:31
*** mattray has quit IRC23:31
*** rustam_ has joined #openstack23:31
*** rnorwood has quit IRC23:31
*** chadh has joined #openstack23:33
*** rnorwood has joined #openstack23:34
*** vidd-away is now known as viidd23:34
*** viidd is now known as vidd23:35
rmkgflags.IllegalFlagValue: flag --vpn_image_id=ami-00000010: invalid literal for long() with base 10: 'ami-00000010'23:36
rmkDocs indicate it wants an image ID in ami notation...23:36
viddrmk, did you copy that from the image list?23:37
rmkI copied it from euca-describe-images23:37
rmkSince the docs specify ami-xxxxxxx23:37
_rfzstill getting the same error23:37
Kiallrmk, i would bet its meant to be an int..#23:39
rmkYeah I would have thought it would want a glance image index id.23:39
rmkWhich is what I gave it now that it failed.23:39
rmkBut all the docs are wrong if that's the case.23:39
*** krow1 has joined #openstack23:40
KiallWell, if they are, I think annegentle looks after the docs :)23:40
*** benner has quit IRC23:41
*** andreas__ has quit IRC23:41
cmasserafKiall, there's a exception in nova-compute:
cmasserafwhen I try to load a instance23:42
*** jkyle has joined #openstack23:42
jkylesometimes i see bugginess in an instance, like it'll auto-shutdown and refuse to reboot. what logs are the best place to look for debug output during that process?23:43
Kiallcmasseraf, thats likely unrelated23:45
Kialli would bet that is the result of asking for the console log of the instance before its active..23:45
KiallBoot a new instance, click nothing else and check the end of the logs..23:45
*** benner has joined #openstack23:47
rmkI guess it doesn't matter what image I specify.23:48
rmkAuth to keystone fails from nova-manage.23:48
Kiallnova-manage doesnt auth against keystone?23:48
*** rnorwood has quit IRC23:49
*** misheska has joined #openstack23:49
rmkUnder normal circumstances, no, but it looks like it is here.23:50
*** coli has joined #openstack23:51
viddnova-manage does not auth against keystone, it auths against the database23:51
rmkCome on.23:51
rmkI obviously know that.23:51
rmkLook at the trace.23:51
viddrmk, Kiall appeard to believe it did23:52
rmkI'm saying there's authentication failing to glance here.23:52
rmkGlance is probably looking for a keystone auth token and not getting one.23:52
rmkAnd it's coming as a result of a nova-manage command.  This log snippet is from nova-manage.log.23:52
*** benner has quit IRC23:53
*** benner has joined #openstack23:54
*** MarkAt2od has joined #openstack23:54
*** lzyeval has joined #openstack23:55
*** dolphm has quit IRC23:56
viddrmk can you paste your glance-api for me?23:56
viddyou can blot out the auth gode [verify it is correct of course]23:57
*** jeremy_ has joined #openstack23:57
*** jeremy has quit IRC23:57
*** jeremy_ is now known as jeremy23:57
*** MarkAtwood has quit IRC23:58
viddrmk, also verify your --keystone_ec2_url=http://x.x.x.x:5000/v2.0/ec2tokens in nova.conf is correct23:59
*** dubsquared has quit IRC23:59
viddand that the region in keystone matches the region in nova.conf23:59
_rfzvidd, did you get vnc working ?23:59

Generated by 2.14.0 by Marius Gedminas - find it at!