Friday, 2011-12-02

*** dailylinux has quit IRC00:00
*** lzyeval has joined #openstack00:01
*** cloudfly has joined #openstack00:05
*** andreas__ has quit IRC00:06
andrewbogottmaplebed:  I've never seen it work, but in theory configuring keystone is easy, just use keystone-manage add, etc.  And then configure swift with swift_auth_server=keystone00:07
andrewbogott(I'm guessing here :) )00:07
maplebedI'm not sure I actually want keystone...00:07
*** whit has quit IRC00:07
andrewbogottMe neither00:07
cloudflywell it's got pluggable modules for auth mechanisms00:08
cloudflythat's pretty nice00:08
*** qazwsx|2 has quit IRC00:08
maplebedyeah - my current setup is using tmpauth.00:10
*** rnorwood has joined #openstack00:10
notmynamemaplebed: I can try to clarify the issue00:14
maplebednotmyname: that would be fantastic!00:14
notmynamemaplebed: what is still confusing?00:14
maplebedI followed the admin docs while setting up my swift cluster.  In those docs, it instructs that I test that it's working by creating an account/user/pass and passing that in with the X-Storage-User header.00:16
notmynamehmm..that doesn't sound right00:16
notmynameah. nm. my mistake00:16
*** lloydde has joined #openstack00:16
notmynameyour are using tempauth?00:16
maplebedA colleague is trying to use the API docs, which all refer to a user and api key, passed in with the X-Auth-User and X-Auth-Key headers.00:16
maplebedI am using tempath (at the moment)00:16
maplebedI am confused about the relationship between the x-storage and x-auth stuff, and haven't found how to create an api key (so can't use the api docs).00:17
maplebedfor reference, the docs I was following and am reading are:00:17
notmynamethe username (X-Auth-User) is the tempauth account:user combo. the api key (X-Auth-Key) is the tempauth password00:18
maplebed, which seems to overlap pretty well with
maplebedand for the dev stuff.00:18
maplebednotmyname: my coworker was reporting that using the account:user and pass didn't work with the x-auth headers; let me try them to confirm..  back in 3.00:19
notmynamemaplebed: the x-auth* vs x-storage* is, as AaronSchulz said, is for legacy clients (since swift started life as a product at rackspace)00:20
notmynamemaplebed: use x-auth-[user|key]00:20
notmynameand x-auth-token00:20
maplebedok, so the simple answer is that x-storage-user and x-storage-pass have been renamed to x-auth-user and x-auth-key.00:21
maplebedalso, my test with curl passed.00:21
maplebed(passed => I get back an x-storage-url and x-storage-token)00:21
maplebedit's interesting that I get back an x-storage-token in addition to x-auth-token (they're the same); the api docs only mention x-auth-token.00:22
maplebednotmyname: thanks; I think I am unconfused for the time being.00:22
*** MarkA2 has quit IRC00:22
*** jakedahn has quit IRC00:23
notmynamemaplebed: ya, x-auth-token and x-storage-token have the same relationship as the other ones. use x-auth-token00:24
maplebedthat paste is very similar to what I have. , though the end of the config URL is AUTH_system rather than AUTH_abc00:25
maplebedwhat is the stuff after the underscore?00:25
*** lloydde has quit IRC00:25
*** lloydde has joined #openstack00:26
uvirtbotNew bug: #898867 in horizon "Launch instance not working after launch instance -> cancel." [Undecided,New]
*** adjohn has joined #openstack00:26
*** pixelbeat has joined #openstack00:27
*** AaronSchulz has quit IRC00:29
*** AaronSchulz has joined #openstack00:30
*** jakedahn has joined #openstack00:30
*** jakedahn has quit IRC00:31
*** tdi has joined #openstack00:31
*** tdi_ has quit IRC00:32
*** rnorwood has quit IRC00:33
*** mattray has joined #openstack00:33
notmynamemaplebed: you mean the "abc" in my example00:33
*** lloydde has quit IRC00:35
*** lloydde has joined #openstack00:35
notmynamemaplebed: a swift account string is made up of 2 parts, separated by the first underscore. the first part ("AUTH" in my example) is the reseller. this allows you to have multiple auth systems for distinct accounts in the cluster without having to ask every auth system about every request. the second part ("abc" in my example) is the unique account string. there's nothing special about the second part. normally we use uuid.uuid4().hex to generate it00:36
maplebedI thought the account part was the testinga portion of the string.00:38
maplebed(so the auth token was account:user)00:38
maplebeds/auth token/x-auth-user header/00:39
maplebedI saw this term 'reseller' in some docs as well.  Can you explain (or point me to docs that explain) what it means?00:40
pweissanyone know if devstack is broken? mine is stuck: + sudo PIP_DOWNLOAD_CACHE=/var/cache/pip pip install --use-mirrors Django==1.3 django-mailer django-nose==0.1.2 django-nose-selenium django-registration==0.7 pycrypto==2.3 sqlalchemy-migrate python-cloudfiles -e git+ -e git+ -e git+
maplebed(I think I understand from your comment that it means you can have something like AUTHONE, AUTHTWO, AUTHTHREE and they'll have distinct users00:42
*** tungvs has joined #openstack00:45
*** pixelbeat has quit IRC00:46
*** whit has joined #openstack00:49
*** adjohn has quit IRC00:50
*** cdub has quit IRC00:51
*** adjohn has joined #openstack00:53
*** dragondm has quit IRC00:54
*** adjohn has quit IRC00:55
*** jaypipes has joined #openstack00:56
*** rustam has joined #openstack00:56
*** MarcMorata has quit IRC00:57
*** AlanClark has quit IRC00:58
*** dotdevops has quit IRC01:02
*** yshh has joined #openstack01:03
*** pringles_ has quit IRC01:04
*** pringles has joined #openstack01:04
*** longhudou has joined #openstack01:06
*** adjohn has joined #openstack01:07
*** adjohn has quit IRC01:07
notmynamemaplebed: the account is testinga:testingb01:07
notmynamemaplebed: the auth token is what the auth server returns and what you pass for subsequent requests01:07
notmynamemaplebed: ya, you are right about the resellers01:08
maplebedthis was the part that confused me: "the second part ("abc" in my example) is the unique account string."01:08
*** livemoon has joined #openstack01:08
notmynamemaplebed: while it's possible to have multiple auth servers without resellers, having resellers allows some efficiencies01:08
*** yshh has quit IRC01:09
*** clopez has joined #openstack01:09
*** yshh has joined #openstack01:09
*** woleium has quit IRC01:09
notmynamemaplebed: for example, at rackspace we have a reseller for the rackspace cloud users and a reseller for jungle disk01:09
*** microtosh has quit IRC01:09
maplebedmakes sense.01:09
notmynamemaplebed: the siwft account string is RESELLER_UNIQUEID01:10
maplebedah, so it's a human-readable vs. internal-unique-id thing?01:10
maplebedI'll use testinga:testingb and internally that turns into AUTH_abc01:10
*** jog0 has joined #openstack01:11
notmynamecorrect. it's the difference in the auth account and the swift cluster account01:13
notmynameone isn't "human readable" compared to the other, they are just different01:13
notmynamethe auth server ties them together01:13
maplebedI was missing the auth account vs. swift account distinction.01:14
maplebedThanks a bunch; your descriptions have been very helpful.01:14
*** vladimir3p has quit IRC01:15
*** pradeep1 has joined #openstack01:17
longhudouI can ping the instance running on the controller node from the controller node ,while from the compute node I can't ping01:17
longhudouhas anybody run into this circumstance before?01:17
*** wilmoore has quit IRC01:17
*** adjohn has joined #openstack01:21
*** adjohn has quit IRC01:22
*** pweiss has quit IRC01:26
*** huslage has joined #openstack01:27
*** adjohn has joined #openstack01:28
*** stuntmachine has joined #openstack01:28
*** adjohn has quit IRC01:28
*** mattray has quit IRC01:28
*** fridim_ has quit IRC01:31
*** dholmberg has quit IRC01:34
*** adjohn has joined #openstack01:38
*** rods has quit IRC01:38
*** bengrue has quit IRC01:40
*** woleium has joined #openstack01:41
*** llang629 has quit IRC01:41
*** Kiall has quit IRC01:46
tungvslonghudou: which networkmanager are you using ?01:46
tungvswhat kind of ipaddress of the vmguest are you using to ping ?01:47
tungvsfixed or floating ?01:48
tungvswhat's compute-node IP ?01:48
*** amdtech has left #openstack01:49
tungvsHas compute-node got any IP ?01:50
*** _adjohn has joined #openstack01:50
*** blakeyeager has quit IRC01:51
longhudouwhat do you mean by that ,I am confused , is a network that instance get ip from? isn't that?01:51
longhudoucan compute-node got an IP?01:51
longhudouhow to ?01:51
*** Ryan_Lane has quit IRC01:52
*** adjohn has quit IRC01:52
*** _adjohn is now known as adjohn01:52
tungvsin my implement, I've assigned my bridge 1 IP in the fixed-range so the host can communicate with the vmguest01:52
*** Ryan_Lane has joined #openstack01:52
*** MarkA2 has joined #openstack01:54
*** songyan has joined #openstack01:54
longhudouwe don't use public IP,01:54
tungvsfixed-rage should be private IP, not public01:55
*** CaptTofu has quit IRC01:55
*** po has quit IRC01:55
longhudouI know that,we are in a LAN,and our IP are 192,we using NAT to get to the public,and I create the fixed IP range with
longhudoudo I need to create the fixed IP with 192?01:57
tungvsin your case, I suggest check to see the compute-host can ping any IP in the, if yes, then dig into the controller iptables to see if it block the compute-host IP, if iptables is OK, then you should use tcpdump to check what happend01:57
*** woleium has quit IRC01:57
*** jog0 has quit IRC01:57
*** woleium has joined #openstack01:57
longhudoufrom the compute-node i can not ping even
tungvsshow me your "ip addr" in the controller and the compute-node01:59
tungvsno, type "ip addr" in your terminal and show us the result02:00
longhudou1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00     inet scope host lo     inet scope link lo     inet6 ::1/128 scope host        valid_lft forever preferred_lft forever 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1                                                         000     lin02:01
longhudou1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00     inet scope host lo     inet scope link lo     inet6 ::1/128 scope host        valid_lft forever preferred_lft forever 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP ql                                                             en 100002:02
tungvsplease use paste.bin02:03
longhudouI am sorry to say that  I don't know paste.bin02:04
longhudouwhere can I find it02:05
longhudouor how to use it02:05
maplebedor or ...  ;)  (there are many around the net)02:05
longhudoucan you find that?02:07
tungvslonghudou: you should assign an IP, like to your compute bridge02:08
*** krow has joined #openstack02:09
tungvsand check whether compute br0 is contain eth0 or not02:09
longhudouhow to assign an IP to compute bridge,I am just a fresh bird02:10
tungvsok, ip addr add  dev br002:11
tungvsand show us "brctl show"02:11
tungvsand "brctl show" of the controller also02:14
tungvsit's short, you could paste it here directly02:14
*** jakedahn has joined #openstack02:14
longhudoubr0             8000.f46d0475fd80       no              eth0                                                         vnet0 virbr0          8000.000000000000       yes02:15
longhudouthis is the controller02:15
tungvsok then02:15
tungvstry to ping again02:15
*** maplebed has quit IRC02:16
longhudouOk,this time I can ping the instance running on the controller node from the compute node02:17
longhudouthanks a lot02:17
longhudoubut there is another problem,@tungvs02:18
longhudouon the controller node I can see an instance running on the compute node ,But I cannot ping02:18
*** jakedahn has quit IRC02:18
tungvsjust use euca-get-console02:20
*** mwhooker has quit IRC02:20
tungvsto check if the vmguest has booted up properly or not02:20
*** huslage has quit IRC02:20
longhudouis shows:::::2011-12-02T02:20:55Z02:21
tungvsanything else ?02:21
longhudouonly this02:21
tungvsthe it cannot boot up02:21
longhudouit's wrong?02:21
*** AaronSchulz has quit IRC02:22
*** clopez has quit IRC02:22
tungvswhat's your compute-node arch, or x64 ?02:22
tungvsI mean 32 or 64 bit ?02:22
*** cp16net has quit IRC02:22
longhudouI install ubuntu11.04 32bit02:22
longhudoudo you mean that? or the hardware?02:23
tungvsand I guess you're using 64 bit image to create vmguest ?02:23
longhudouoh,no,the  host system and the vmguest is the same02:24
tungvswhat hypervisor are you using ?02:25
*** spiffxp has quit IRC02:25
tungvsare your hardware support that ?02:25
*** bencherian has quit IRC02:25
longhudoumaybe it is caused by the wrong image?02:26
tungvsdo a "ps -ef | grep kvm" in your compute-node02:26
longhudouI create it manually02:26
tungvscan you see any process ?02:26
*** bencherian has joined #openstack02:26
longhudouroot      1025     2  0 Nov30 ?        00:00:00 [kvm-irqfd-clean]02:26
tungvsanything else ?02:26
*** pradeep1 has quit IRC02:26
longhudou116       4132     1  0 09:24 ?        00:00:36 /usr/bin/kvm -S -M pc-0.14 -enable-kvm02:27
longhudouroot      4133     2  0 09:24 ?        00:00:00 [kvm-pit-wq]02:27
*** jakedahn has joined #openstack02:27
*** rustam has quit IRC02:27
tungvsjust paste it all02:27
*** bencherian has quit IRC02:28
longhudouOk, it is on the page360002:29
tungvsuse VNC to connect to your vmguest02:29
tungvsmaybe something's worng with your image02:29
tungvsyou should download an ec2 image to create vmguest first02:30
tungvsto make sure your nova setup is ok02:30
longhudoui think something is wrong with my image,too02:31
*** reed has quit IRC02:32
*** andrewbogott has quit IRC02:32
*** stuntmachine has quit IRC02:32
*** Kiall has joined #openstack02:34
longhudouI create it manually ,and in the installation process, I create more than one partition(root,swap,for example),and after that, i did not adjust the image,does openstack only support ext4 image?02:34
larissalonghudou: Error: "tungvs" is not a valid command.02:34
*** redconnection has joined #openstack02:34
*** misha has joined #openstack02:34
*** misha is now known as Guest8712702:34
*** misheska has quit IRC02:35
*** clauden_ has quit IRC02:37
tungvsI suggest you stick with the guide if you don't know what you're doing. I think you have to adjust the image because nova will mount the image as root partition, so the image should be read as a partition, hence must be adjusted02:37
tungvscan anyone please verify that for us ?02:38
*** stuntmachine has joined #openstack02:38
longhudouOk, then i will download ec2 image to have a try02:38
*** cmagina has quit IRC02:38
longhudouthanks very much02:38
*** camm has quit IRC02:39
*** clauden_ has joined #openstack02:40
*** stuntmachine has quit IRC02:40
*** Guest87127 has quit IRC02:44
*** osier has joined #openstack02:52
*** adjohn has quit IRC02:58
*** AaronSchulz has joined #openstack03:00
*** camm has joined #openstack03:02
*** AaronSchulz has quit IRC03:04
*** agrajag has quit IRC03:08
*** agrajag has joined #openstack03:09
*** packet__ has quit IRC03:15
*** andrewbogott has joined #openstack03:18
*** janpy has joined #openstack03:22
*** redconnection has quit IRC03:26
*** misheska has joined #openstack03:28
*** misheska has quit IRC03:28
*** andrewbogott has quit IRC03:32
*** hadrian has quit IRC03:32
*** rnorwood has joined #openstack03:38
*** dolphm has joined #openstack03:40
*** cmagina has joined #openstack03:42
*** wilmoore has joined #openstack03:43
*** agrajag has quit IRC03:45
*** agrajag has joined #openstack03:45
*** koolhead17 has quit IRC03:45
uvirtbotNew bug: #898935 in keystone "bin/keystone doesn't respect port settings" [Undecided,In progress]
*** rnorwood has left #openstack03:46
*** bombay has joined #openstack03:48
*** Ryan_Lane has quit IRC03:49
*** stuntmachine has joined #openstack03:52
*** rsampaio has joined #openstack03:52
*** robbiew has quit IRC04:03
*** jdurgin has quit IRC04:03
*** misheska has joined #openstack04:03
*** misheska has quit IRC04:04
*** janpy has quit IRC04:04
*** janpy has joined #openstack04:05
tungvsI've created an instance_type with swap space about 512 MB. But when I boot vmguest up and do "fdisk -l" as root, it shows /dev/vda,vdb,vdc dont contain valid partition table. And "swapon -s" show nothing, means no swap. Can anyone explain this ?04:10
*** stuntmachine has quit IRC04:11
*** dolphm has quit IRC04:14
*** CaptTofu has joined #openstack04:15
*** packet_ has quit IRC04:15
*** jj0hns0n_ has quit IRC04:19
*** ldlework has joined #openstack04:19
*** jj0hns0n has joined #openstack04:20
*** scottjg has joined #openstack04:21
*** lborda has joined #openstack04:21
*** jj0hns0n has quit IRC04:22
*** vidd has joined #openstack04:33
*** Rajaram has joined #openstack04:33
*** vidd has quit IRC04:34
*** vidd has joined #openstack04:34
*** jfluhmann has joined #openstack04:38
*** jakedahn has quit IRC04:39
*** nati2 has quit IRC04:39
*** janpy has quit IRC04:43
*** vidd-away has joined #openstack04:44
*** vidd-away has joined #openstack04:44
*** vidd has quit IRC04:45
*** jakedahn has joined #openstack04:51
*** Ryan_Lane has joined #openstack04:51
*** jakedahn has quit IRC04:51
*** supriya has joined #openstack04:53
*** adjohn has joined #openstack04:54
*** adjohn has quit IRC04:59
*** Ryan_Lane is now known as Ryan_Lane|away05:01
*** lborda has quit IRC05:05
*** PeteDaGuru has quit IRC05:15
*** ldlework has quit IRC05:19
*** koolhead17 has joined #openstack05:22
*** pradeep has joined #openstack05:26
*** koolhead17 has quit IRC05:31
*** wilmoore has quit IRC05:32
*** pradeep has quit IRC05:34
*** dpippenger has quit IRC05:34
*** tungvs has quit IRC05:35
*** tungvs has joined #openstack05:42
*** mjfork has quit IRC05:44
*** rsampaio has quit IRC05:46
*** rsampaio has joined #openstack05:49
*** pradeep1 has joined #openstack05:49
*** pradeep1 has joined #openstack05:50
*** ghanima has quit IRC05:56
*** jward has joined #openstack05:56
*** tmichael has quit IRC05:59
*** jward has joined #openstack05:59
*** tmichael has joined #openstack06:00
*** Frodent has quit IRC06:06
*** janpy has joined #openstack06:06
*** reed has joined #openstack06:08
*** rsampaio has quit IRC06:10
*** rsampaio has joined #openstack06:12
*** jward has quit IRC06:13
*** supriya has quit IRC06:14
*** fulcrum has joined #openstack06:16
*** jward has joined #openstack06:17
*** supriya has joined #openstack06:18
*** reed has quit IRC06:19
*** nerens has joined #openstack06:24
*** zefry has joined #openstack06:26
zefryhello all06:26
zefrycan you help me about cloud-publish-image i386 serverfinal.img bucket106:27
zefrythe result is : failed to check for existing manifest06:28
zefryanyone, can you help me?06:28
*** arBmind has joined #openstack06:33
*** tungvs has quit IRC06:33
*** tungvs has joined #openstack06:33
*** fulcrum has quit IRC06:35
*** objectiveous has joined #openstack06:36
*** rsampaio has quit IRC06:40
zefryanyone, can you help me?06:51
tungvszefry: I guess you should check if the glance image service runs or not, the config in nova.conf about glance IP:port, and the sql databse which glance connects to06:56
*** mindpixel has joined #openstack07:07
*** wilmoore has joined #openstack07:10
*** pradeep1 has quit IRC07:16
*** jakedahn has joined #openstack07:20
*** jakedahn_ has joined #openstack07:22
*** jakedahn has quit IRC07:25
*** jakedahn_ is now known as jakedahn07:25
*** wilmoore has quit IRC07:25
*** jakedahn has quit IRC07:28
*** pradeep1 has joined #openstack07:32
*** javiF has joined #openstack07:33
*** matiu has joined #openstack07:34
tungvsI've created an instance_type with swap space about 512 MB. But when I boot vmguest up, run "swapon -s" show nothing, means no swap. Can anyone explain this ?07:36
*** mgoldmann has joined #openstack07:38
*** woleium has quit IRC07:44
tungvsanyone knows what's in the /var/lib/nova/instances/_base ? I find its size is huge07:47
*** tmichael has quit IRC07:47
*** tmichael has joined #openstack07:48
*** jedi4ever has joined #openstack07:49
*** lloydde has quit IRC07:49
*** miclorb_ has quit IRC07:50
*** kaigan_ has joined #openstack07:50
*** marcuz has joined #openstack07:51
marcuzgood morning07:51
uvirtbotNew bug: #898983 in horizon "cancel button on launch instance window really hard to see" [Undecided,New]
*** arBmind has quit IRC07:54
*** adiantum has joined #openstack07:54
*** pixelbeat has joined #openstack07:57
*** foexle has joined #openstack07:58
*** adiantum has quit IRC08:01
foexleahoi @all08:01
*** GheRivero has quit IRC08:03
tungvsMy nova-compute sometimes just hangup. When I make nova-compute restart, it hangs at the log line "Connecting to libvirt: qemu:///system.."08:04
tungvsAny idea ?08:04
*** GheRivero has joined #openstack08:06
*** reidrac has joined #openstack08:07
*** TheOsprey has joined #openstack08:07
*** binbash_ has quit IRC08:11
*** binbash_ has joined #openstack08:12
*** krow has quit IRC08:13
*** sloop has quit IRC08:14
*** Ramonster has joined #openstack08:14
*** sloop has joined #openstack08:15
*** sloop has joined #openstack08:15
*** jfluhmann has quit IRC08:18
*** sloop has quit IRC08:24
*** sloop has joined #openstack08:26
*** javiF has quit IRC08:27
*** guigui has joined #openstack08:32
*** sloop has quit IRC08:39
*** popux has joined #openstack08:47
*** vdo has joined #openstack08:51
*** apurva has joined #openstack08:52
*** oubiwann has quit IRC08:52
*** sloop has joined #openstack08:52
apurvaneed help getting invalid literal for int() with base 10: '' error while running euca-authorize -P icmp -t -1:-1 default command08:54
*** Razique has joined #openstack08:54
*** uksysadmin has joined #openstack08:56
*** janpy has quit IRC08:58
*** dobber has joined #openstack08:59
*** PotHix has joined #openstack09:08
Sander^workI found an bug in the docs.. condradiction theese two places: and
Sander^workin the endpointTemplates i'm unsure if I should include the tentant_id at the end of the url for nova09:25
Sander^workOne of the docs say one thing, the other say something else... Anyone know what's right?09:26
*** nacx has joined #openstack09:28
*** PiotrSikora has quit IRC09:30
apurvaany one can please help me on euca-authorize command error09:31
*** PiotrSikora has joined #openstack09:32
Sander^workAnyone around that have set up keystone?09:36
*** scottjg has quit IRC09:38
marcuzSander^work: agree, it generates an exception on line 9 of "keystone/manage/", isn't it?09:40
zefryhello @all09:43
*** fulcrum has joined #openstack09:44
zefryI have trouble with uploading to openstack09:44
Sander^workmarcuz, I don't know where to look for that exception.. in which log files?09:44
*** songyan has quit IRC09:45
Sander^workmarcuz, do you know what the right value of nova endpointTemplates is?09:45
zefrycommand is : cloud-publish-image i386 serverfinal.img bucket109:45
zefryand result is: failed to check for existing manifest09:46
zefrydo you have a solutions for it?09:46
marcuzSander^work: can't go in detail, i was working on it a couple of days ago and i figured out it's breaking when creating the user, i misread your question, sorry :)09:46
zefrymarcuz: can you help me?? :(09:47
Sander^workmarcuz, I'm unable to log in to the openstack dashboard.. just gives me same page back.09:48
*** rustam has joined #openstack09:57
*** apurva has quit IRC10:04
*** lzyeval has quit IRC10:07
*** katkee has joined #openstack10:08
*** yshh has quit IRC10:10
tungvszefry: I guess you should check if the glance image service runs or not, the config in nova.conf about glance IP:port, and the sql databse which glance connects to10:11
*** clopez has joined #openstack10:13
zefrytungvs: glance is running,10:14
Raziquezefry: it's a custom image ?10:14
*** adiantum has joined #openstack10:15
uvirtbotNew bug: #899056 in openstack-ci "python-novaclient needs milestone-proposed Jenkins jobs" [High,Confirmed]
zefryRazique: I don't understand about custom image..10:17
Raziquezefry:where did u get the image from ?10:18
zefryfrom release.ubuntu.com10:18
*** mcclurmc has quit IRC10:18
Raziqueok that's the issue here10:19
Raziquemmm wait10:19
zefryRazique: okey, thank you before :)10:19
*** mcclurmc has joined #openstack10:20
Raziquezefry: do you have a man for it ?10:20
RaziqueI can't find it10:20
*** BasTichelaar has joined #openstack10:21
Raziquein fact I think it looks for "cloudified" images10:21
Raziqueeg. splitted kernel, ramdisk and image10:21
*** livemoon has quit IRC10:21
Raziquesince you are trying to upload an ISO (non "cloudified" image :D ) the command can't find the manifest (xml file that describes the structure of your image10:22
Raziqueyou would want to use these images :10:22
Raziquethat will work for sure10:22
zefryokey I'm try ..10:22
foexleheyho Razique how are you :)10:23
Raziquefoexle: :D10:23
Raziquehi dude10:23
Raziquewe missed each other 2 days ago!10:23
RaziqueI was up 'til 2 am10:23
zefryRazique: thanks for sharing :)10:24
foexleyeah i know :( .... i sleepd 30 mins before youe came on :/10:24
Raziquehehe tonight then ?10:24
Raziquezefry: sure10:24
*** javiF has joined #openstack10:24
foexleRazique: oh today i'll go out with friends10:25
foexleRazique: because my friends says every day ... "Hey heiko where are you, why you don't have time" and so on ;)10:25
*** rods has joined #openstack10:26
BasTichelaar hey guys10:26
Raziquehehe yah10:26
foexleRazique: so should we do next week .... if you have time just say me ... EY heiko lets go hahahahaha10:26
BasTichelaaranyone has setup a multi-zone cluster?10:26
Raziquewe geek don't use "Social Life" as PPA often10:26
uvirtbotfoexle: Error: "^" is not a valid command.10:26
zefryRazique: what is a good format to run? tar.gz or .img ??10:26
Raziquehey BasTichelaar :)10:26
Raziquezefry: tar.gz for cloud-utils10:27
BasTichelaarRazique: hey, how are you doing?10:27
zefryRazique: thanks :)10:27
RaziqueBasTichelaar: playing with swap provisionning for instances10:27
RaziqueSwapless instances are scary10:27
BasTichelaarRazique: swap provisioning? whats that?10:27
BasTichelaarRazique: ah ok10:27
Raziqueso I'm working on a way to do swap provisionning based on the actual RAM10:27
Raziquenothing much but I've something that works here10:28
BasTichelaarRazique: shouldnt be to difficult, are you gonna hack it into openstack?10:28
RaziqueBasTichelaar: no, I don't know Python much10:28
Raziqueso I created a bash script meant to be run from rc.local atm10:28
BasTichelaarRazique: its quite simple, there is a file where the commands are run for a created instance10:29
BasTichelaaryou can modify them to include a swap partition10:29
Raziqueoh great!10:29
BasTichelaarits just cfdisk commands10:29
Raziquethanks i'll look for that10:29
Raziquewhat about you ? still working on multi-zones ?10:29
BasTichelaarRazique: yes, but I have a very simple use case, that doesnt seem to be implemented10:30
BasTichelaarI have two zones, one in amsterdam and one in zurich10:30
BasTichelaarand I wanna be able to specify the zone where an instance should live10:30
BasTichelaarbut it seems I just have to connect to a different API if I'm correct10:30
BasTichelaarbut then projects are spread across multiple zones, but there is no global knowledge10:31
Raziqueyah i think you are rigth BasTichelaar10:31
BasTichelaaranyway, I have opened a question for it10:31
BasTichelaarhope sandy will answer10:31
*** zefry has quit IRC10:32
colibastichelaar: look at multiclusterzones10:32
RaziqueI'm also waiting for instance's resizing under KVM10:32
Raziquethat would be cool10:32
coliyou need two zones as child of parent zone10:32
colithen talk to api for the parent zone10:32
BasTichelaarcoli: yes, but you cannot specify the zone manually right?10:32
BasTichelaarcoli: I did implement it, but found no way to do what I want10:33
BasTichelaarcoli: or should I fire up instances in child zones, and for the overview connect to the parent zone?10:33
colibastichelaar: hmm... look into api code and see if there is a zone parameter for create (DON't look into documentation which is way outdated)10:34
tungvsRazique: talking about swap in instance, I've created an instance_type including swap. When I boot vmguest up, I see an extra /dev/vdb, but it contains no swap partition, can you explain the proper way to have swap in vmguest ?10:34
BasTichelaarcoli: well, there is one for availability zones in python-novaclient10:35
BasTichelaarbut not for the multi-cluster zones10:35
colibastichelaar: I would expect one to be there, otherwise it would be no point for zones except only for internal usage.10:35
colion the other hand, what you are saying that talking to two api's is not that bad10:35
colido yo uuse dashboard or your own frontend ?10:36
BasTichelaarcoli: no, thats what I thought10:36
BasTichelaarcoli: we're not using a frontend, but the python-novaclient api10:36
colibastichelarr: I will be playing with it this weekend if I find time :-)10:36
BasTichelaarcoli: cool :)10:36
coliso look into nova-api code as well and see if there is zone parameter for create and if it's used later on :-)10:37
BasTichelaaryes, I'm now reading through the api code10:37
Raziquetungvs: yah... it only provides a disk for swap10:38
Raziquebut doesn't enable it10:38
Raziqueso format it and swapon-it10:39
tungvsRazique: sure I will, tks10:39
colianybody used RADOS (rbd) as storage for nova ?10:39
Raziquethat's also why my swap creation is scripted by now10:39
tungvsRazique: script ? How can you script to do something inside a newly created instance ? I'm interested to know.10:41
RaziqueI put a ref to my swap script on /Etc/rc.local10:43
Raziqueon boot it is executed10:43
Raziqueas the metadatas are retrieved10:43
*** livemoon has joined #openstack10:44
BasTichelaarRazique: idea: provide the size of the swap in the metadata :)10:44
Raziquehi livemoon :)10:44
RaziqueBasTichelaar: how ?10:44
Raziquewhat's up man ?10:44
foexlehi livemoon10:44
BasTichelaarRazique: by setting the metada and reading them out (they get written to /meta.js inside the instance)10:45
RaziqueBasTichelaar: what would be the gain ?10:45
Raziquemore of a "structural" better way ?10:45
Raziqueso swap should be seen as a metadata also ?10:45
BasTichelaarRazique: well, then you can at least specify the size of the swap on instance creation10:46
BasTichelaarRazique: dont know if its needed for your usecase10:46
RaziqueBasTichelaar: yah, my script does that too10:46
Raziquebut I'm goan update my base images10:46
Raziqueso they include the script and a first swapfile10:47
livemoonyour new script about swap?10:47
Raziqueso on boot in many cases, we won't loose time by creating a swapfile10:47
*** popux has quit IRC10:47
Raziquelivemoon: yah10:47
livemoonprovide another swap?10:47
Raziquelivemoon: nope; the only one10:47
Raziquetoday instances are swapless10:47
*** zigo has joined #openstack10:47
*** longhudou has quit IRC10:47
Raziqueso that script creates a swap based on the flavor10:48
Raziqueand set a 1gb thresholf10:48
livemoonwhen I create rhel image,if there is no swap, installation will failed10:48
livemoonDoes your script be pushed into your github?10:49
tungvslivemoon: swap partition can be made when installing RHEL, and be tripped off using losetup after that for saing space10:49
*** MarkA2 has quit IRC10:49
livemoontunavs: tks, I will try10:50
tungvsRazique: maybe I'm the only noob here, but I want to ask how to retrieve flavor through "metadata" in the instance ?10:51
tungvsand what is "metadata"  btw ?10:51
foexlean example for metadatas are the used key-pair10:55
tungvsoh I see, that's the "curl" step, right10:56
tungvsok, I will dig deeper, tks foexle10:57
*** zigo-_- has joined #openstack10:57
Raziquetungvs: yup it's an Amazon stuff10:58
Raziquethe api server contains info about the newly created instance10:59
Raziquethat one connects to it and retrive insfos10:59
Raziquepublic keys mainly10:59
tungvsRazique: is there any document about it ?10:59
Raziqueso you can ssh to the instance after it booted, without the pain of setting password10:59
livemoonapi server?10:59
*** saju_m has joined #openstack11:00
tungvsthat's great, Razique11:00
tungvshave to go for dinner, good to see you guys.11:00
*** zigo has quit IRC11:01
*** zigo-_- is now known as zigo11:01
livemoonwe have the same timezone maybe11:02
Raziqueyou welcome tungvs11:02
Raziquelater :)11:02
*** saju_m has quit IRC11:02
*** saju_m has joined #openstack11:03
*** coli_ has joined #openstack11:03
*** coli has quit IRC11:03
*** saju_m has quit IRC11:03
*** saju_m has joined #openstack11:04
*** saju_m has joined #openstack11:05
Raziqueow man11:05
RaziqueI did nothing this morning11:05
Raziquejust been lazy on internet11:05
livemoonI am busy today11:06
livemooncoding to monitor my VM11:07
*** saju_m has quit IRC11:07
Raziquehow livemoon  ?11:07
*** saju_m has joined #openstack11:07
livemoonusing libvirt11:07
*** tungvs has quit IRC11:07
*** saju_m has quit IRC11:08
livemoonI just start to learn11:08
*** saju_m has joined #openstack11:08
livemoonI think your work about metadata is useful for me11:08
*** ahasenack has joined #openstack11:11
*** katkee has quit IRC11:11
Raziquei'll see how I could well integrate that extra step11:16
livemoonteach me please11:18
RaziqueBasTichelaar: here suggested to tweak the api data to be provided11:19
Raziqueif you can get the ram, you can also get the swap11:19
*** GheRivero has quit IRC11:23
*** GheRivero has joined #openstack11:25
*** livemoon has left #openstack11:26
*** titan has joined #openstack11:32
*** popux has joined #openstack11:32
*** miclorb_ has joined #openstack11:33
*** fulcrum has quit IRC11:35
zykes-notmyname: ?11:37
titanHi guys, here comes an unsolicited questions: does openstack have a zone wide DNS resolution solution or something similar in the blueprints? At the moment, we rely on dnsmasq to read local file to achieve dns resolution. However, should more than one dnsmasq running in different servers, there seems to be no way to cross servers to do name resolution11:40
titanthx in advance, guys11:40
foexlehey Razique do you use swift ?11:44
Raziquefoexle: not yet :(11:44
foexleah ok11:44
*** cyberdo has joined #openstack11:44
foexlebut a question .... i found a resolved high critical bug ... this one was resolved in nov. but i cant see a merge in the stable ubuntu packages11:45
cyberdoquick questions: does openstack support automatic migration using local files? Can it use LVM as these local files?11:46
*** saju_m has joined #openstack11:46
foexlenormally should the package maintainer resolve this bug in the package version too11:46
Raziquecyberdo: what do you mean by "local file" ?11:47
cyberdoRazique: non-shared storage... storage is held on the virtualization nodes11:47
*** miclorb_ has quit IRC11:48
Raziqueyah both ways11:49
Raziquelive migration or block storage migration11:50
cyberdoour nodes use very little disk (often about 10G per server), so I rather keep the disk locally... Also it reduces the impact of failures (having a SAN breaking down is not funny)11:50
cyberdoooh.. nice11:50
cyberdoalso: this using xen?11:50
Raziquecyberdo: I think it's supported with Xen11:50
cyberdo(I saw that some features differed between different hosts)11:50
Raziqueit's the kvm way11:50
cyberdothanks a lot11:51
Raziquebasically nova-manage vm migrate --ec2-id --dest11:51
Raziquenova-manage vm live_migration11:51
Raziqueand nova-manage vm block_migration11:51
titanHi guys, here comes an unsolicited questions: does openstack have a zone wide DNS resolution solution or something similar in the blueprints? At the moment, we rely on dnsmasq to read local file to achieve dns resolution. However, should one dnsmasq running in different servers in multi_host mode, there seems to be no way to cross servers to do name resolution11:54
cyberdoRazique: 'Instances: Instance can be migrated with ISCSI/AoE based volumes'  ;o/11:55
Raziquecyberdo: what about lvm ?11:56
Raziqueyah it is said, but never really tried yet11:56
*** nijaba has quit IRC11:57
Raziquetitan: zykes- here is the man11:58
zykes-Razique: ?11:58
Raziquezykes-: titan has a DNS question :p11:59
RaziqueHi guys, here comes an unsolicited questions: does openstack have a zone wide DNS resolution solution or something similar in the blueprints? At the moment, we rely on dnsmasq to read local file to achieve dns resolution. However, should one dnsmasq running in different servers in multi_host mode, there seems to be no way to cross servers to do name resolution11:59
titanzykes: yep, I have one11:59
*** dendro-afk is now known as dendrobates11:59
*** zz_bonzay is now known as bonzay12:00
titanRazique: thx12:00
cyberdoRazique: I suppose I can still look into it... otherwise there might be some distributed iSCSI-server out there12:00
zykes-titan: dns howo ?12:00
Raziquecyberdo: I need to validate migration with iscsi12:00
RaziqueI was afraid the I/O part12:01
zykes-yeah, there was an idea on "dnsaas"12:01
zykes-but I haven't had the time to do anything on it12:01
zykes-atm i'm working on a thing with Swift instead12:01
*** nijaba has joined #openstack12:01
Raziquewhat happens if there are ongoing writes on the instance while you perform the migration12:01
*** bonzay is now known as zz_bonzay12:01
titanzykes: is that okay to say that there is no zone wide DNS service at the moment?12:02
zykes-titan: think so yeah12:03
titanzykes: cool. thx112:04
zykes-i'm trying to front swift as a archival service for media files12:04
*** nijaba has quit IRC12:04
*** saju_m has quit IRC12:05
BasTichelaartitan: as far as I know, openstack does nothing with dns12:09
BasTichelaarhmm, didnt scroll to the bottom :)12:09
zykes-check that out12:10
*** nijaba has joined #openstack12:17
*** nijaba has joined #openstack12:17
*** supriya has quit IRC12:19
*** ben_duyujie has joined #openstack12:21
*** shawn_ is now known as shawn12:29
*** mcclurmc has quit IRC12:31
*** mcclurmc has joined #openstack12:33
*** ben_duyujie has quit IRC12:35
coli_razique: what do you mean by writes ?12:36
*** bsza has joined #openstack12:37
*** ben_duyujie has joined #openstack12:37
uvirtbotNew bug: #899116 in keystone "The wrong keystone module is used when building docs" [Undecided,In progress]
*** po has joined #openstack12:42
*** n8 has joined #openstack12:42
*** n8 is now known as Guest317112:43
*** pradeep1 has quit IRC12:52
*** ben_duyujie has left #openstack12:52
*** markvoelker has joined #openstack12:56
*** livemoon has joined #openstack12:58
*** PeteDaGuru has joined #openstack13:04
*** mcclurmc has quit IRC13:04
*** mcclurmc has joined #openstack13:05
zykes-notmyname: here ?13:08
*** yshh has joined #openstack13:11
*** PeteDaGuru has quit IRC13:12
*** yshh has joined #openstack13:16
*** lts has joined #openstack13:17
*** dolphm has joined #openstack13:17
*** egant has quit IRC13:18
notmynamezykes-: morning13:20
*** osier has quit IRC13:21
*** javiF has quit IRC13:21
Spirilisok question- When using "nova-manage network create ...", it looks like it completes without error (and /var/log/nova/nova-manage.log looks good), but no entries are added to the DB ... what's the first thing I should check?13:23
SpirilisThe /etc/nova/nova.conf file has the correct --sql_connection string13:23
*** lloydde has joined #openstack13:23
*** abecc has joined #openstack13:24
Spirilisand it looks like the DB user specified in there does have GRANT ALL PRIVILEGES access to that db13:24
Spirilisfyi this is using Ubuntu's natty repo copy of openstack, minus the keystone piece which I git'd13:25
SpirilisOneiric repo, not natty13:27
Spirilisusing ubuntu 11.10 here13:27
Spiriliscan't keep track of their crazy names ;)13:27
foexleanyone knows how i can change password of an user with keystone-manage ?13:27
zykes-gawd this company and opensource13:30
zykes-funny thing13:30
*** Razique has left #openstack13:32
*** mcclurmc has quit IRC13:33
*** mcclurmc has joined #openstack13:34
*** mjfork has joined #openstack13:34
*** lorin1 has joined #openstack13:34
*** Razique has joined #openstack13:35
zykes-notmyname: can you go from having 2 nodes with swift with 2 zones on each13:36
zykes-to having 4 nodes + with 1 zone pr box or rack ?13:36
Spiriliszykes-: yes, if you are trying to move from a "2 zones/node" to "1 zone/node" config it might be prudent to add the new nodes first, rebalance the rings once or twice, let it sync up, then remove one of the extra zones from the original 2 servers, rebalance/let it resync, then re-add that disk under the same zone as the one already on the box13:38
*** Tristani3D has joined #openstack13:38
Tristani3DI got a question13:39
Tristani3DI am following the Starter Guide but I am running into an issue13:39
Tristani3Don the service iscsitarget restart part13:39
Tristani3DI receive the error: FATAL: Module iscsi_targt not found13:40
Tristani3Dany ideas?13:40
Tristani3Dps. I have installed isxcsitarget-dkms13:40
mjforkTristani3D: on Ubunutu?13:40
mjforkhmm - guessing here - do you have nova-voume installed?13:41
mjforkcan you pastebin the full trace13:41
Tristani3Dnova-volume is indead part of the nova install section13:41
Tristani3Dill pastebin the output, 1sec.13:43
notmynamezykes-: what Spirilis said13:44
larissaTristani3D: Error: "mjfork" is not a valid command.13:47
zykes-but i guess i'll never the the light of swift here anyways, people here want a more conventional storage option for a media archive :p13:48
notmynamezykes-: sounds frustrating13:48
zykes-notmyname: yeah, kindof13:48
mjforkTristani3D: so it is NOT openstack throwing this error?13:48
notmynamezykes-: any arguments I can add that would help?13:49
zykes-notmyname: i dunno, a person here wants it to be cheap so :p13:49
*** mnabil has joined #openstack13:49
zykes-and i guess 30k $ isn't ckeap enough for a 24tb deployment13:49
Tristani3Dmjfork: I have no clue13:49
mjforkso, modprobe is failing oot ;ppls ;ole13:50
mjforkwoops, modprobe is failing it looks like13:50
notmynamezykes-: 1) scales more than conventional storage 2) since the replicas/self-healing nature you don't have to have separate hardware for backups 3) can use commodity hardware (saving money)13:50
zykes-notmyname: but what about off site ?13:51
zykes-like in case of a flood or so :p13:51
Tristani3Dmjfork: well, strange, I have followed every line of the guide ;)13:51
mjforkyeah, unknown to me.  what version of ubunutu?13:51
zykes-notmyname: the solution we're going for now is 2*24tb dell boxes with windows + veeam backup13:52
zykes-the 2nd box will replicate the first box over a wan link13:52
notmynamezykes-: ya, you could either have a 2nd cluster and sync or you could do something like 4 zones (2 in each location). with 3 replicas you are guaranteed at least one copy in both locations13:52
mjforkTristani3D: i don't know, but that should fix you up13:52
*** dprince has joined #openstack13:52
Tristani3Dmjfork: Ill give it a go, and let you know. Thx13:52
*** zz_bonzay is now known as bonzay13:52
zykes-notmyname: but say for trunked objects then, will 1 replica be a full copy of all chunks ?13:52
*** bonzay is now known as zz_bonzay13:53
Tristani3Dmjfork: meh, is 404 :(13:54
*** mcclurmc has quit IRC13:54
*** GheRivero has quit IRC13:54
notmynamezykes-: no, each chunk will be in 2 of the 4 locations. if you always have 4 zones, you will always have at least 1 copy of each entity in both places. you'd have to be aggressive about hardware failures though (eg HDD with the only copy in location A goes down and may not be restored before location B goes down)13:54
notmynamezykes-: s/2 of the 4/3 of the 4/13:55
*** mcclurmc has joined #openstack13:55
zykes-notmyname: doesn't each chunk have 3 replicas ? :|13:55
*** yshh has quit IRC13:55
zykes-what you mean by being aggressive then ?13:55
*** objectiveous has quit IRC13:56
*** katkee has joined #openstack13:57
mjforkTristani3D: what is that? i think you just need to run those 3 commands13:59
mjforkTristani3D: apt-get then m-a then modprobe iscsitarget13:59
mjforkjust skip down to installing the packages14:00
*** mnabil has quit IRC14:01
Tristani3Dmjfork: Done, seems to work now14:04
Raziqueargh (nova.compute.manager): TRACE: Stderr: 'ERROR: trying to add VLAN #100 to IF -:None:-  error: No such device\n'14:04
*** katkee has quit IRC14:04
Tristani3Dmjfork: iscsi-source seems to be the trick and missing package14:04
Raziquebridge_interface=eth0 in nvoa.conf and NULL into db ; any hint ?14:04
Tristani3Dmjfork: might be a good one to add this in the docs14:05
*** hadrian has joined #openstack14:05
*** AlanClark has joined #openstack14:05
Raziquemjfork: can you confirm on my comment ?14:06
Raziqueif it's working, i'll update the doc so14:06
zykes-can't the damned prices go down for disks soon14:07
Raziquezykes-: yh :/14:07
mjforkRazique: what commments?14:07
mjforkTristani3D: what guide where you following?14:07
zykes-i might as well give up this fight, useless :p14:07
Raziquemjfork: weren't you talking about nova-volume setup issue14:07
Raziqueregarding the iscsi part14:08
Tristani3Dmjfork: "Openstack Compute Starter Guide" Nov 11, 201114:08
livemoonhi, Razique14:08
mjforklooks like Razique documented the bug that Tristani3D ran into at that link14:09
livemoonhow to use your swap scripte when booting an instance?14:09
mjforkand the fix14:09
foexlelivemoon: you need to add this script in the startup area14:10
livemoonfoexle: how to use in novaclient command?14:10
foexlenovaclient? .... not with novaclient14:11
notmynamezykes-: just that in my example there is a slight risk of data loss if the one copy in one location is lost at the same time the entire other location is lost14:11
foexleyou need change you image startup14:11
RaziqueTristani3D: ok great14:11
RaziqueI'll update the doc14:11
Raziquei'm looking into the source file and reduce the line size14:11
livemoonfoexle: teach me14:11
*** rsampaio has joined #openstack14:12
zykes-if the disk prices had been what they where earlier14:12
livemoonadd into rc.local?14:12
zykes-one simple swift deployment would cost only 15520$14:12
Raziqueyah :/14:12
Raziquezykes-: ramdisk ?14:12
zykes-Razique: no, 2tb sata prices14:12
Glaceelol ramdisk is still expensive :)14:13
foexlelivemoon: which distro ?14:13
Glaceeeven for vm load14:13
zykes-if you work out of the price increase % and subtract that from todays disk prices14:13
zykes-and take todays server prices without disks14:13
notmynamezykes-: how big is that deployment?14:13
livemoonRazique: I think nova-volume docs is old14:13
livemoonif we use 11.10, do not use iscsitarget14:14
zykes-notmyname: 4 boxes of 24 tb notmyname14:14
Tristani3DRazique: btw, the lines are not in the starter guide its seems14:14
Tristani3DRazique: but in the OpenStack Compute Administration Manual14:15
Tristani3DRazique: page 10314:15
Glaceehave anyone tried the new Seagate 3Tb drive that came out last week? Seems pretty neat14:15
Tristani3DRazique: and the image on page 85 is bugged as well, way to big for an A4 sized page14:16
coli_tristani3D: docs are for developers, however your best documentation is in the source code,14:16
GlaceezykesL: with which brand did you finally go with for your seervers?14:16
*** rsampaio has quit IRC14:16
coli_tristani3D: then correct that bugged page :-) it's open source14:16
RaziqueTristani3D: don't hesistate to open doc bugs for things you notice14:17
RaziqueI regularly track doc bugs and try to fix. dispatch em14:17
zykes-Glacee: Dell R51514:17
zykes-came out the cheapest actually14:17
Tristani3Dcoli: Razique: Thanks, im new to this. No idea how and where to edit :)14:18
Glaceeok cool14:18
Glaceecheapeast than SM?14:18
zykes-but i'm getting sold out to a customer anyways, tired of not getting ideas through here so might as well help someone else14:18
Raziqueclick "Report a bug"14:18
Raziqueand we'll see it14:18
Glaceehehe ok14:18
*** bryguy has quit IRC14:18
*** danishman has joined #openstack14:18
Tristani3DRazique: ok14:18
*** mrjazzcat has joined #openstack14:19
*** nelson1234 has quit IRC14:21
*** bryguy has joined #openstack14:21
*** doude has joined #openstack14:21
doudeHi stackers14:22
Raziqueoi doude  :)14:23
doudeWho use the Diablo release with libvirt KVM on Ubuntu 11.10 ?14:23
Tristani3Dcoli_: Razique: filled those two bugs, thx14:23
*** sgg has joined #openstack14:23
doudeRazique: hello14:23
RaziqueTristani3D: perfect14:23
*** lvaughn has joined #openstack14:24
*** dolphm has quit IRC14:24
doudeI get libvirt freeze evry 24 hours. Do you met that problem ?14:25
zykes-notmyname: how come ?14:25
notmynamezykes-: quick calculations from newegg and dell websites: r815 (2800) + 2 supermicro JBODs (4400) + 90 2TB HDDs (90*255) / 90000 (marketing GB) / 4 (marketing to billable conversion) = hardware corst of $.08 per billable GB14:25
*** mrjazzcat has left #openstack14:26
notmynamezykes-: how come what?14:26
Raziquedoude: any error ,14:26
*** gnu111 has joined #openstack14:27
doudeRazique: strange, I've got it on 3 differents hosts14:27
Raziqueanyone have met ERROR: trying to add VLAN #100 to IF -:None:-  error: No such device\n ?14:27
doudeDo you use Ubuntu 11.10 ?14:27
Raziqueit's fricking weird :/14:27
doudeand libvirt version ?14:28
*** lborda has joined #openstack14:28
*** rsampaio has joined #openstack14:28
Raziquein fact i'm amazed, since It was working fine14:28
doudemine libvirt 0.9.214:29
Raziquethe subtility here is that my node uses eth0 while the controller eth114:29
Raziquedoude: can you check syslog ?14:29
zykes-notmyname: why the 90*255 ?14:29
doudeNothing in syslog14:30
notmynamezykes-: HDD cost on newegg14:30
notmynameto fill up the jbods14:30
zykes-yeah, but then comes the hw cost in addition or ?14:30
Raziquedoude: nova-compute.log ?14:30
doudeI just found this in libvirt logs :
zykes-notmyname: what disks is that ?14:32
doudeRazique: nova-compute wait a response from libvirt and the compute service is frozen14:32
foexledoude: is you database inconsistent ?14:33
doudeif I restart the libvirtd daemon, the compute restart itself14:33
doudefoexle: no the db stay clear14:33
doudeevrything work fine after I restarted the libvirt daemon14:34
zykes-why you need 90 of them ?14:34
doudesome use libvirt 0.9.2 ?14:34
*** dolphm has joined #openstack14:34
foexlehow many instances are running ?14:34
notmynamezykes-: 2 45-slot JBDS14:34
zykes-that's not the dell r815 ?14:35
foexledoude: i'm using 0.9.214:35
doudefoexle: and no problem ?14:35
foexledoude: nope .... no problems like this ;)14:36
notmynamezykes-: says $280014:36
foexledo you have checkd your hdd's with smart or controller bios ?14:36
Tristani3DI have a general question, am I right that Openstack doens't rely on SAN storage but distribute its storage to all disks in every single node instead?14:36
zykes-what's the r815 for ?14:37
notmynamezykes-: the controller for the 2 supermicro JBODS14:37
ashpTristani3D: I think, and I'm just an idiot that wants to use openstack eventually myself, you can do it either way.  SAN or distributed filesystem of some kind.14:37
notmynamegotta run. be back in about an hour (ish)14:38
Glaceenotmyname: do you think its less expensive.. buy a small 1u server + jobd.. or buying the monster 4U server with 36 slots14:38
Tristani3Dashp: Right, without the use of a SAN its going to be much cheaper..14:38
doudefoexle: Ok thanks. I'll try to found a way to reproduce that14:38
*** yshh has joined #openstack14:39
*** gavri has joined #openstack14:39
foexledoude: yeah sorry :)14:39
ashpTristani3D: That's one reason I'm interested - obtaining shared storage here is a nightmare14:41
zykes-notmyname: we don't want jbods14:42
*** ryker has joined #openstack14:42
zykes-sorry notmyname i meant SM14:42
*** oubiwann has joined #openstack14:48
*** darthflatus has joined #openstack14:49
*** eugene has joined #openstack14:51
Tristani3Dashp: yeah, i like to concept of using all the disk in all nodes instead of one (or multiple) storage instance14:51
*** eugene is now known as Guest3610214:51
*** Guest36102 is now known as eugenes14:52
*** rustam has quit IRC14:54
*** ryker has quit IRC14:55
*** ryker has joined #openstack14:55
zykes-notmyname: does RS today do cross site replicating ?14:56
*** aliguori has joined #openstack14:58
*** rustam_ has joined #openstack14:59
*** dgags has joined #openstack15:05
*** mjfork has quit IRC15:06
*** katkee has joined #openstack15:06
uvirtbotNew bug: #899186 in openstack-manuals "Line size bugged page 103" [Undecided,New]
Raziqueregarding nova-flags15:08
Raziquethey either reside in database or nova.conf ?15:08
uvirtbotNew bug: #899189 in openstack-manuals "Image to large at page 85 of OpenStack Compute Administration Manual" [Undecided,New]
*** katkee has quit IRC15:11
btorchzykes-: he is probably on the way to the office, but we don't do cross DC replication if that's what you mean15:11
*** mjfork has joined #openstack15:11
zykes-btorch: so one can't have data across datacenters ?15:11
btorchzykes-: Not saying that you can't, swift has container sync and there is nothing really stopping one from having a zone in different location. You need to make sure you have a good pipe to reach that location15:12
btorchzykes-: like  a dedicated link for example15:13
*** imsplitbit has joined #openstack15:14
*** code_franco has joined #openstack15:15
*** ryker_ has joined #openstack15:16
Tristani3DOk I am having this next issue15:18
Tristani3Droot@ubuntu:/home/localadmin# nova-manage db sync15:18
Tristani3DCommand failed, please check log for more info15:18
Tristani3DAny idea, which logs?15:18
*** clauden_ has quit IRC15:18
*** clauden has joined #openstack15:18
*** ryker has quit IRC15:19
*** ryker_ is now known as ryker15:19
*** Rajaram has quit IRC15:20
*** clauden__ has joined #openstack15:21
*** clauden has quit IRC15:21
*** clauden__ is now known as clauden15:21
*** eugenes has left #openstack15:21
gnu111Tristani3D: check nova-api.log, I think. or nova-manage.log.15:22
Glaceezykes: You can have multiple datacenter in the same cluster/ring15:23
GlaceeYou need to makre sure that you have a low latency line between both sites and that you seperate your failure zones properly15:23
Glaceeit will slow down swift.. if your link is the bottleneck15:24
Glaceedont do it over the internet for an example :015:25
*** dubsquared has joined #openstack15:25
*** deshantm_laptop has joined #openstack15:25
Tristani3Dgnu111: thx, I see (nova): TRACE: OperationalError: (OperationalError) (1045, "Access denied for user 'novadbadmin'@'' (using password: YES)") None None15:25
*** cmagina has quit IRC15:25
*** GheRivero has joined #openstack15:25
*** kbringard has joined #openstack15:25
Tristani3Dgnu111: might have been a misconfiguration somewhere15:25
*** cmagina has joined #openstack15:25
gnu111Tristani3D: is the mysql db started? also check the firewall. make sure the compute node can talk to the db.15:26
Tristani3Dgnu111: mysql     2331  0.0  2.4 174400 24980 ?        Ssl  06:52   0:00 /usr/sbin/mysqld15:26
Tristani3Dgnu111: mysql is running15:27
dubsquaredcan you log into the db with the username and password specified in the nova.conf?15:27
*** qazwsx has joined #openstack15:27
gnu111is the compute node or the controller?15:28
*** primeministerp has joined #openstack15:28
*** CaptTofu has quit IRC15:28
*** sgg has quit IRC15:28
Tristani3Dgnu111: controller, im following the compute starter guide section "server 1"15:28
*** ldlework has joined #openstack15:29
Tristani3Ddubsquared: ERROR 1045 (28000): Access denied for user 'novadbadmin'@'localhost' (using password: YES)15:30
notmynamezykes-: sorry. my math was a little wrong earlier. I should have divided by 180000 (for total marketing GB rather than 90000)15:30
notmynamezykes-: (2800 + 4400 + (90*255)) / 180000. / 4 = $0.041875 / GB15:30
Sander^workAnyone around that have got horizon with keystone working?15:30
gnu111Tristani3D: there you go. did you do something like this for the db: GRANT ALL PRIVILEGES ON $DB_NAME.* TO $DB_USER IDENTIFIED BY '$DB_PASS ?15:31
Tristani3Dgnu111: mysql -uroot -pmygreatsecret -e "SET PASSWORD FOR 'novadbadmin'@'%' =15:31
Raziquesoren: here ?15:31
sorenRazique: Kinda. Gimme 5 minutes.15:31
Tristani3Dgnu111: direct copy of the docs, but due to the enter it fails to execute proparly15:31
Raziquesoren: sure thanks15:31
Tristani3Dgnu111: root@ubuntu:/home/localadmin# nova-manage db sync15:32
Tristani3D  MigrateDeprecationWarning15:32
Tristani3Dworks now15:32
*** katkee has joined #openstack15:32
Tristani3Dor not?15:32
*** reed has joined #openstack15:34
*** ldleworker has joined #openstack15:34
*** ldlework has quit IRC15:35
notmynamezykes-: oops. it was pointed out to me that I should have multiplied by 4 rather than divide by 415:35
notmynamezykes-: that makes it $.67 per GB15:35
notmynamemuch closer (but still lower) than your specs15:35
*** katkee has quit IRC15:37
sorenRazique: 'sup?15:38
Raziquefine and you ,15:39
Razique? erf :D15:39
Raziqueok I've a a issue here with that file15:40
*** livemoon has left #openstack15:40
gnu111notmyname: do you know if cyberduck works with the tempauth setup in swift? I have only been using the command line so far.15:40
Raziqueit get the bridge_interface from the database15:40
notmynamegnu111: it should (but I haven't done it myself)15:41
RaziqueI've tried to add a flag 'bridge_interface'" so it uses nova.conf15:41
coli_razique: what networking mode are you working in ?15:41
Raziquemy issue here is that i've a node which doesn't use the same interface as the others15:41
Raziquecoli_: vlan15:41
Raziqueso a flag would override the default eth015:41
coli_razique: why woyld you use bridge_interface then for vlan mode ?15:41
*** dolphm has quit IRC15:41
*** jaypipes has quit IRC15:42
Raziquecoli_: you don't need it ?15:42
RaziqueMy vlan doesn't up15:42
coli_razique: will check, just a sec15:42
RaziqueStderr: 'ERROR: trying to add VLAN #100 to IF -:None:-  error: No such device\n'15:42
*** primeministerp has quit IRC15:42
RaziqueNone because the database doesn't contain a bridge_interface15:42
RaziqueI'm using diablo15:43
coli_razique: just checked to be sure15:43
Raziquethe bridge is up on the vlan interface15:43
coli_you need to use vlan_interface, on that interface it's going to create vlan interfaces, remember that vlan mode create one vlan for each tenant15:43
Raziqueyup I've the flag vlan_interface :)15:44
*** wilmoore has joined #openstack15:44
Raziquebut I looked into the code (the file I pasted)15:44
Raziquethe vlan uses the bridge_interface15:44
Razique            _execute('vconfig', 'add', bridge_interface,15:44
Raziqueand bridge_interface is get from the database15:44
coli_if you set vlan_interface to eth3 and tenant is got vlan 155 assigned then nova-network will create vlan155 as vlan interface of eth1 and bridge br155 with vlan155 and interfaces from instances15:45
coli_razique: for flat and flatdhcp15:45
Raziquewhat about vlan mode the ?15:45
Raziquewhy does it use a bridge ?15:46
coli_just looking into the code, what line was it ?15:46
Razique892 on my paste15:46
*** dolphm has joined #openstack15:46
Raziquelook around line 860 also15:46
Razique                          network['bridge_interface'],15:46
Raziqueit would be ideal if we defined a new flag15:47
Raziqueeg flags.DEFINE_string('bridge_interface','eth0', 'blah')15:47
Raziquebut I don't know how to "enable it "15:47
*** Tristani3D has quit IRC15:48
kbringardthe flags object gets read in, and handles reading it from the config file15:48
*** jj0hns0n has joined #openstack15:48
kbringardso if you added that flag, you would enable it by just setting —bridge_interface in the nova.conf15:48
Raziqueoi kbringard :)15:48
Raziquekbringard: I tried, no luck15:48
kbringardhmmm, I'd have to look at it… I have not added a new flag since before the cactus release, hehe15:49
*** vicelow has joined #openstack15:49
kbringardso maybe the mechanism changed15:49
RaziqueI also replaced "                          network['bridge_interface'],15:49
Raziqueby "FLAGS.bridge_interface"15:49
vicelowhello all15:49
Raziquebut no luck15:49
Raziquehey vicelow15:50
coli_flags.DEFINE_string('vlan_interface', None,                     'vlans will bridge into this interface if set')15:50
coli_from manager.py15:50
Raziquecoli_: oh, so i's not the "bridge_interface" fielsd ?15:50
Raziquefield into database*15:51
Raziquebecause It's set into my nova.conf --vlan_interface=eth215:51
uvirtbotNew bug: #899225 in nova "EC2API: rescue instance can't work" [Undecided,In progress]
coli_the vlan's should be created on eth2 and bridge into brNNN where NNN is the vlan id15:51
Raziquecoli_: yah15:52
Raziqueweird then my flag is not read :/15:52
Raziquewhile it was under Cactus15:52
RaziqueI didn't noticed it earlier cuz cactus had set the vlan properly15:53
Raziqueand if it exists, it's not recreated15:53
coli_did you restart nova-network ?15:53
Raziquebut here I rebooted the lab and saw that15:53
Raziqueand nova-compute15:54
*** rnorwood has joined #openstack15:54
coli_check your networks table if it's got correct vlan's bridges set15:54
*** popux has quit IRC15:54
Raziquecoli_: which field into networks table ?15:54
*** dragondm has joined #openstack15:54
*** dragondm has quit IRC15:55
*** dragondm has joined #openstack15:55
coli_razique: all of the table if it looks right15:55
coli_razique: has all the paremeters defined and explained at the begining of the file15:55
*** wilmoore has quit IRC15:57
Raziquecoli_: the manager deals with the network table doesn't it ?15:57
Raziqueit looks like you can't have different interfaces for vlans15:57
Raziqueeg one server eth3 another one eth215:57
coli_looks fine to me (your paste)15:57
Raziquesince I don't see where the flag bridge_interface doesn't exist15:57
coli_what do you mean you cannot have different interfaces for vlans ?15:57
Raziquecoli_: nova-network on server 1 uses eth015:58
Raziquenova-compute on server 2 uses eth315:58
coli_on each nova-network host you can have different bridge15:58
RaziqueI don't use multi_host15:58
coli_do you have nova-network on each compute host ?15:58
Raziquenope one nova-network atm15:58
coli_so just one nova-network providing info for all compute hosts ?15:58
Raziqueand on every node a nova-compute only15:59
coli_in that situation you are right and you can only have one interface :-)15:59
Raziqueok so If I install nova-network on the compute node15:59
Raziquethe flag "bridge_interface" and vlan_interface will work ?15:59
Raziquecoli_: oh I think I get it16:00
*** robbiew has joined #openstack16:00
*** pradeep has joined #openstack16:00
coli_you need just vlan_interface if you have nova-network on each host16:00
coli_I would also strongly suggest nova-api on each host16:00
Raziquecoli_: why is that ?16:00
Raziquecoli_: ok leme try :D16:01
*** adjohn has joined #openstack16:01
coli_razique: look how packets travel to when starting instance and you will see :-)16:02
*** uksysadmin has quit IRC16:02
*** kaigan_ has quit IRC16:02
coli_razique: if you are going to have separate interface just for that traffic then fine you can have just one nova-api, however if you use public interface then you have a problem :-)16:02
*** javiF has joined #openstack16:03
*** imsplitbit has quit IRC16:03
*** javiF has joined #openstack16:03
*** mindpixel has quit IRC16:03
*** cp16net has joined #openstack16:04
Raziquecoli_: the api is not publically exposed16:05
Raziquecoli_: you are the man here16:05
Raziqueit's working16:05
Raziquethe reason why I didn't set up multi_host before16:05
Raziqueis bcause I found out that it was avoiding a live migration Cactus -> Diablo16:06
Raziquelive : without service interruption16:06
*** cctrieloff has joined #openstack16:06
*** reidrac has left #openstack16:06
coli_razique: it;s not to do with publicly exposed api but with snat and dnat which is done on the compute host.16:06
Raziqueah ?16:06
coli_currenlty on remote host, how is starting instance communication with api ( ?16:07
Raziqueactually the compute node uses the prerouting route in order to access nova-api16:08
*** helfrez has joined #openstack16:09
Raziquecoli_: in fact maybe I did it wrong the last time16:09
Raziquewith a running cloud, can I enable/ disable multi_hosting ?16:10
coli_razique: it's using preroute and then that packet is routed via public interface to nova-api or thorugh other ?16:10
Raziquedoes multi_hosting is mandatory when I've one nova-network per nova-compute also16:10
Raziquelet me check16:10
*** freeflyi1g has joined #openstack16:10
coli_razique: I would think so (no idea, just guessing)16:10
Raziquecoli_:   101  6060 DNAT       tcp  --  *      *       tcp dpt:80 to:
Raziqueok so it's a local prerouting16:12
*** bencherian has joined #openstack16:12
Raziquethat hist the nova-api directly16:12
Raziquecoli_: that's the thing I don't get in fact regarding diablo16:13
coli_and then it uses public interface to get to ?16:13
Raziquewhy was I able to have two vlan interfaces without issue16:13
*** freeflying has quit IRC16:13
Raziquewhile with Diablo I need a nova-network per nova-compute16:13
Raziquetwo vlan interfaces = --vlan_interface that was not the same on every compute node16:14
Raziquecoli_: the private interfaces is on the same range16:15
mjforkRazique: i just got a single machine running nova-network with VLANs and many compute nodes, is this what you are trying to do16:15
*** misheska has joined #openstack16:15
Raziquemjfork: ok so the issue here is that with cactus, I was able to have one nova-network with it's vlan_interface=eth016:15
mjforkoh man, no idea on cactus16:15
Raziqueand on every node vlan_interface=eth1, eth2 etc...16:16
coli_razique: then you don't have a problem ;-) if it was routed via public interface then it would have been snated and 'cause nova-api provides meta-data based on source ip it wouldn't find it ;-(16:16
*** primeministerp has joined #openstack16:16
Raziquecoli_: yah I see what you mean16:16
Raziquebut my vlan issue remain weird though16:16
*** guigui has quit IRC16:17
*** f-yang has joined #openstack16:17
*** fred_ has joined #openstack16:17
*** primeministerp has quit IRC16:17
Raziquemjfork: i'll do extra test16:17
*** fred_ is now known as Guest699816:17
*** Guest6998 has left #openstack16:18
Raziqueeg remove nova-network, for instance16:18
Raziqueand see how routring works16:18
*** reed has quit IRC16:18
*** rnirmal has joined #openstack16:18
Raziqueerf I just saw the isntance can't reach the metadata server now16:19
Raziqueok i'll go home and debuf that16:19
Raziquelaters guys16:19
*** Razique has quit IRC16:19
*** axisys has joined #openstack16:20
*** yshh has quit IRC16:21
*** adiantum has quit IRC16:24
*** debo-os has joined #openstack16:25
*** spiffxp has joined #openstack16:25
*** TheOsprey has quit IRC16:25
*** martines has quit IRC16:26
*** jaypipes has joined #openstack16:31
*** imsplitbit has joined #openstack16:32
*** juddm has joined #openstack16:36
*** pradeep has quit IRC16:37
*** dobber has quit IRC16:38
*** clopez has quit IRC16:38
*** oubiwann has quit IRC16:40
*** martines has joined #openstack16:41
*** oubiwann has joined #openstack16:43
*** livemoon has joined #openstack16:43
*** wilmoore has joined #openstack16:44
*** maplebed has joined #openstack16:46
*** katkee has joined #openstack16:47
*** misheska has quit IRC16:49
*** koolhead17 has joined #openstack16:53
*** dolphm has quit IRC16:53
*** marcuz has quit IRC16:54
*** clopez has joined #openstack16:54
*** negronjl has quit IRC16:59
*** misheska has joined #openstack16:59
*** misheska has quit IRC16:59
*** negronjl has joined #openstack16:59
*** bencherian has quit IRC17:01
*** anotherjesse has joined #openstack17:01
*** Ramonster has quit IRC17:03
*** nerens has quit IRC17:03
*** vladimir3p has joined #openstack17:04
*** Razique has joined #openstack17:04
RaziquePing from the apple store17:04
Raziquedjee what are theses places =D17:04
Raziquemjfork: still around ?17:04
*** foexle has quit IRC17:05
*** ShawnB has joined #openstack17:05
zykes-notmyname: around still ?17:05
*** ShawnB has left #openstack17:06
notmynamezykes-: yes17:07
mjforkRazique: yes17:08
Raziquemjfork: are you using diablo ?17:08
zykes-notmyname: has anyone run swift on solaris ?17:09
notmynamezykes-: nexenta has mentioned some stuff on the mailing list (Caitlin Bestler). I'm not sure if anyone from there is here in IRC17:10
Raziquemjfork: so you have several nodes but only  one nova network installed ?17:10
RaziqueStill trying to figure out how the vlan_interface is used17:10
mjforkVLAN interface is the interface on node used to establish the VLAN17:11
Raziquedo you have a different vlan_interface per node ?17:11
Raziqueyah I know :D17:11
*** juddm has quit IRC17:12
mjforkall of my vlans need to be created on eth117:12
Raziquemjfork: ah ok17:12
mjforkmy cloud controlelr creates all VLANs on eth117:12
mjforkthat is the network device with access to all ym compute nodes17:12
Raziqueyayah, that's why it's working17:12
mjforkwhat do you have?17:13
RaziqueI've different vlan_interface per node :D17:13
Raziqueand that is why without multi_host mode, the vlan cannot be created17:13
*** llang629 has joined #openstack17:14
mjforkso each node has a dedicated interface on teh cconteroller?17:14
Raziqueand it doesn't seem there have been changes on that part except the multi_host stuff17:14
Raziquemjfork: yah, but it's never the same17:14
Raziqueon the nova-network, it's eth017:14
Raziqueeth1 sorry17:15
Raziquewhile it's eth2 on all nodes17:15
mjforkwell, just set vlan_interface on each of the nodes to eth217:15
mjforkthat is used by nova-compute to setup vlans17:16
*** Corollary has joined #openstack17:16
mjforkon nova-network it would be eth017:16
Raziquemjfork: that is what I use to have with Cactus :)17:17
Raziqueit wa working fine17:17
CorollaryHi I'm running a XenServer pool right now (non openstack) with 80 nodes, however XS technically only supports pools of size 16 for shared storage and live migration.. does OpenStack support larger pools for VM migration and shared storage?17:17
Raziquebut since I migrated to Diablo, it's no longer working ; I think it has something to do with the multi_host moe17:17
*** katkee has quit IRC17:18
Raziquethe only solution I quicklly tried was to instal nova-network on the node, so it uses the vlan_interface flag17:18
*** TheOsprey has joined #openstack17:18
*** llang629 has left #openstack17:19
mjforklooking at mine17:20
*** swill has joined #openstack17:20
Raziqueit looks like the bridge is used after as Vish statew17:21
mjforkmy nova-compute log says vlan_interface eth117:21
mjforkand looks to be creating the VLAN on that itnerface17:21
Raziquemjfork: yup17:22
Raziquemjfork: is it production ?17:23
mjforkdefine production?17:23
RaziqueI mean is your architecture here is production  or your lab ?17:24
*** cleverdevil has joined #openstack17:24
mjforkit is in my lab for playing17:24
mjforknot a production setup.17:24
Raziquewould you try something for me ?17:24
Raziqueenable multi_host mode and see if it's breaks running network17:24
mjforki tired it17:24
Razique(while you have running instances)17:24
mjforkit did17:25
mjforki had all kinds of problems17:25
*** CaptTofu has joined #openstack17:25
mjforkmy original try had multi_host with nova-network on every node17:25
mjforki couldn't get it working quite right17:25
mjforkso i reverted back17:25
Raziquemjfork: have u seen that link
mjforkthat is waht i tried17:25
Raziquethanks for that feedback then17:25
Raziquethat means I'm screwed if the nodes reboot =D17:26
mjforkyou must be doing production?17:26
Raziquei did migration on production without multi_host mode17:26
Raziquerunning fine17:26
Raziqueon my lab, I rebooted and that how I figured the issue17:26
Raziquemaybe If I can figure how to live enable multi_host mode i'm safe17:27
*** PotHix has quit IRC17:27
Raziquebut based and your feedback, and my brief tests, it won't be that easy hehe17:27
*** nacx has quit IRC17:28
*** cleverdevil has left #openstack17:28
*** juddm has joined #openstack17:29
Raziquegrrr that Apple Store makes me sick =D17:29
Raziqueeverything is so polished and shiny17:29
*** helfrez has quit IRC17:29
*** clauden has quit IRC17:30
mjforkyes, i had issues, i would find someplace to paly with17:30
*** clauden has joined #openstack17:30
Raziquemjfork: here is what 'ill try : 1- remove nova-network and multi_host from the node17:30
Raziquesince the vlan and the bridge won't be down17:30
Raziquei'll be able to keep the instances runninn without routing issue17:31
Raziquethen 'ill reinstall nova-network and enable multi_host and see where it breaks the network17:31
Kiallthe issue is probably going to the the default gateway inside the instances..17:32
Kiallthey will all point at 1 IP.. the wrong IP17:32
swilli am writing an auth middleware for swift right now and I am confused about something. according to the documentation (, there should be a first request with User and Key and then they will be given a token.  How is the auth middleware supposed to handle this?  none of the examples I have seen show that.17:32
Kiallwormman mentioned he made some edits so all nova-network nodes used the same fixed_ip..17:33
KiallNo idea how though17:33
RaziqueKiall: I could add a pre routing rule then ?17:33
Raziquefor the "transition" phase17:33
notmynameswill: the auth middleware doesn't handle the token part (well, it could, like swauth). the middleware is responsible for checking to see if a token if valid for the given request17:33
swillnotmyname: who creates the token then?  it has to be my middleware if i am integrating with cloudstack because they will not give a token otherwise.17:35
*** reed has joined #openstack17:35
*** arBmind has joined #openstack17:36
swillnotmyname: so should I just handle the request to /v1.0 differently and create a token based on the X-Auth-User and X-Auth-Key?17:36
notmynameswill: your auth server. the middleware piece is the glue between the auth server and the proxy server. your auth middleware /could/ do everything (like swauth does, but it's really just there to glue the parts together17:36
notmynameswill: exactly17:36
* notmyname lunch time17:36
RaziqueKiall: mjfork
swillnotmyname: ok, i will see what i can do.17:37
mjforkyes, saw that.17:37
KiallRazique: what about it?17:37
mjforkand running nova_network + nova_api will alllow to work17:37
mjforkwhich is what that specific things was aimed at17:37
*** yehudasa has joined #openstack17:38
yehudasavishy: did you see our changes to the blueprint, it was pretty much rewritten last week?17:38
swillnotmyname: it looks like tempauth also handles this piece as well.  i will review those two and see what i can come up with.17:39
Raziquemjfork: ow ok ; so the Ha also concurs nova-api ?17:39
Raziqueif so that would be the solution thenè17:39
mjforkRazique: what do you mean?17:40
*** Ryan_Lane|away has quit IRC17:40
*** dysinger has joined #openstack17:40
*** MarcMorata has joined #openstack17:40
mjfork"so the Ha also concurs nova-api ?"17:41
Raziquewell in fact setting up multi_host mode seems to break metadata routing17:43
Raziquesetting up a local nova-api would resolve the issue maybe17:43
*** andrewbogott has joined #openstack17:44
Raziquei'll try that on the lab tonight ; and I'll give you a feedback. is having two nova-api could be dangerous within the same zone ?17:45
*** dgags has quit IRC17:45
*** dotdevops has joined #openstack17:46
*** rustam_ has quit IRC17:47
*** vdo has quit IRC17:47
*** rustam has joined #openstack17:48
*** vipul has joined #openstack17:48
Raziquegtg thanks mjfork17:49
*** Razique has quit IRC17:49
*** rustam_ has joined #openstack17:52
*** nati2 has joined #openstack17:53
*** rustam has quit IRC17:55
*** rustam_ has quit IRC18:01
*** cctrieloff has quit IRC18:02
*** nerens has joined #openstack18:03
*** rods has quit IRC18:03
*** nerens has quit IRC18:04
*** AlanClark has quit IRC18:06
*** GheRivero has quit IRC18:08
*** jdurgin has joined #openstack18:10
*** livemoon has quit IRC18:11
uvirtbotNew bug: #899299 in nova "nova-manage expext to register images with int id" [Undecided,New]
uvirtbotNew bug: #899302 in nova "Context goes None intermittently due to calls to get_admin_context method" [Undecided,New]
uvirtbotNew bug: #899298 in openstack-qa "Can not get tenant quota from /{version}/{tenantID}/os-quota-sets" [Undecided,New]
*** BasTichelaar has quit IRC18:13
*** AlanClark has joined #openstack18:14
*** dysinger has quit IRC18:14
*** gyee has joined #openstack18:15
*** juddm has quit IRC18:16
*** bencherian has joined #openstack18:17
*** AaronSchulz has joined #openstack18:18
*** juddm has joined #openstack18:21
*** dgags has joined #openstack18:22
*** clopez has quit IRC18:25
*** jog0 has joined #openstack18:25
*** jog0 has left #openstack18:25
*** tmichael has quit IRC18:26
*** MarcMorata has quit IRC18:27
*** crescendo has joined #openstack18:27
*** dysinger has joined #openstack18:28
*** bencherian has quit IRC18:28
*** Corollary has quit IRC18:29
*** AlanClark has quit IRC18:29
*** dysinger has quit IRC18:29
*** woleium has joined #openstack18:31
uvirtbotNew bug: #899310 in tempest "Self parameter in rand_name(  is not needed" [Undecided,Invalid]
*** Ryan_Lane has joined #openstack18:34
*** lloydde has quit IRC18:34
*** ollie1 has left #openstack18:40
*** cdub has joined #openstack18:42
*** dragondm has quit IRC18:45
*** tmichael has joined #openstack18:46
*** bencherian has joined #openstack18:52
*** danishman has quit IRC18:53
vicelowShould I use special hardware in order to use Openstack?18:57
*** bengrue has joined #openstack18:58
*** nerens has joined #openstack18:58
Kiallvicelow: that really depends on your needs..18:59
vicelowthanks Kiall,19:00
KiallBut - Basically anything that can run VMs will work..19:00
Kiall(ie a VT enabled CPU, and lots of RAM)19:00
vicelowInteresting!... I will read more about this19:01
*** dolphm has joined #openstack19:04
notmynameswill: ya, tempauth does too19:05
swillnotmyname: ya, i think i understand how it works a little better now.  i will implement it as i think it should be done and i am sure the community will review my implementation once i have somethign working.19:06
notmynameswill: cloudstack is citrix, right?19:07
*** tmichael has quit IRC19:07
swillnotmyname: it is now, they recently bought it.  i believe it was an opensource project before they bought it.19:08
swillnotmyname: it is still opensource, but citrix has a dev team working on it a lot right now.19:08
*** dgags has quit IRC19:13
*** bencherian has quit IRC19:15
*** MarkAtwood has joined #openstack19:16
*** anotherjesse has quit IRC19:17
uvirtbotNew bug: #899330 in glance "create WADL for image-api" [Undecided,New]
*** reed has quit IRC19:18
*** zigo has quit IRC19:19
*** dysinger has joined #openstack19:21
*** dpippenger has joined #openstack19:25
*** bcwaldon has joined #openstack19:26
notmynamegnu111: nice tweet :-)19:27
gnu111notmyname: thanks!. I have been thinking about some use cases for swift. storing old log files we rarely use is a serious headache.19:28
juddmgnu111: but why would you need 3+ replicas of those files?19:29
notmynamegnu111: we use swift for that. we use the swift clusters to store logs that the swift clusters generate. it works really nicely19:29
pyis it keystone part of diablo itself or plugin?19:29
notmynamepy: keystone is a separate openstack project, but it's only officially a core project after diablo19:29
gnu111juddm: I know. I had the same thought. i probably don't need all the replicas.19:29
*** tmichael has joined #openstack19:30
notmyname(although there are backports and dependencies)19:30
pynotmyname: aha tnx19:30
pyis it still plain text for passwords in diablo notmyname ?19:30
pylike it was in B19:30
notmynamepy: you've now reached the limit of my keystone knowledge :-)19:30
notmynamepy: you should ask dolphm19:31
notmynamegnu111: are you looking for more use cases or just thinking about a few in some more detail?19:32
*** reed has joined #openstack19:32
*** bcwaldon has quit IRC19:33
*** egant has joined #openstack19:33
gnu111notmyname: Looking for use cases that could be useful for us here. Most of our users (academic researchers) are used to Lustre type systems for their data. So I am just trying to think how they could take advantage of swift.19:34
zykes-gnu111: store static files19:40
zykes-for one19:40
*** mwhooker has joined #openstack19:46
*** reed has quit IRC19:50
*** afmtab has joined #openstack19:51
*** nati2 has quit IRC19:58
*** javiF has quit IRC20:00
juddmgnu111: massive file share20:00
*** reed has joined #openstack20:02
vicelowIs node.js supported in OpenStack object storage in the web service API ?20:03
*** AlanClark has joined #openstack20:03
*** rnorwood1 has joined #openstack20:05
*** rnorwood has quit IRC20:06
dolphmpy: passwords salting & hashing was added at the last minute in diablo (and is available in stable/diablo) ... we use passlib, fyi20:09
zykes-vicelow: what you mean by supported ?20:09
*** GheRivero has joined #openstack20:10
*** bcwaldon has joined #openstack20:12
*** codepoet has joined #openstack20:13
vicelowI just read that the currently supported API bindings for object storage are php, python, java, c# and ruby20:14
*** sloop has quit IRC20:16
*** sloop has joined #openstack20:17
notmynamevicelow: those are the language bindings that are for rackspace cloud files (which is a superset of swift--ie it includes rackspace-specific cdn features)20:17
*** ziyadb has joined #openstack20:17
codepoetI've been trying to get a test install of openstack up following the starter guide, but i'm having trouble actually starting an instance... everything looks okay but when i try to start an instance it sticks at pending, and nothing ever gets placed into /var/lib/nova/instances ... can anyone point me to where i need to look for the problem?20:18
jaypipessoren: ping20:20
vicelownotmyname: but coud I develop a node.js app that can get files from swift storage?20:20
jaypipessoren: could you make me a driver/admin for the tempest project please? I need to create milestones/series and assign bugs and stuff..20:20
notmynamevicelow: I suppose so. the API for swift is HTTP REST. the language bindings simply wrap up that language's http libs and provide some helper functions. if you have the ability to do arbitrary requests and set headers on them in node.js, then you can easily make a node.js binding20:24
*** dragondm has joined #openstack20:24
*** dragondm has quit IRC20:25
vicelowthanks notmyname20:25
*** renuka has joined #openstack20:25
*** dragondm has joined #openstack20:25
*** renuka has quit IRC20:26
*** afmtab has quit IRC20:27
*** bencherian has joined #openstack20:29
*** bencherian has quit IRC20:30
*** debo-os has quit IRC20:33
*** AlanClark has quit IRC20:34
*** rnorwood1 has quit IRC20:34
*** ameade has quit IRC20:34
*** ameade has joined #openstack20:34
*** rnorwood has joined #openstack20:34
*** localhost has quit IRC20:36
uvirtbotNew bug: #899374 in tempest "NameError: global name 'exceptions' is not defined when build times out" [Low,Triaged]
*** localhost has joined #openstack20:37
*** arBmind has quit IRC20:37
Glaceenotmyname: What do you think of this setup? 4U server nodes.. with 36 3TB 7.2k drives.. instead of using head node+ 45 bay JBOD?20:38
*** mahmoh has quit IRC20:38
GlaceeIf my calculations are right.. it gives around 9disk per U.. vs 10 with head node + 2 servers.. but I think at smaller case it would make more sense to use the 4U.. it should not take a big hit if a node goed down20:40
Glaceehead node + 2 jbods*20:40
*** bencherian has joined #openstack20:43
*** dysinger has quit IRC20:44
notmynameGlacee: ya, it depends also on your needs (DC power availability, etc). but that would work find20:45
uvirtbotNew bug: #899383 in tempest "Out of the box, tempest does not run cleanly, needs config file" [Medium,In progress]
Glaceenotmyname: ok thanks20:46
*** dprince has quit IRC21:10
*** bcwaldon has quit IRC21:11
*** bcwaldon has joined #openstack21:11
*** lorin1 has quit IRC21:11
*** markvoelker has quit IRC21:11
*** rnorwood has quit IRC21:11
*** sniperd has joined #openstack21:12
sniperdHi All, having trouble getting over setting up permissions on swift - Im getting an error "ERROR: swift.cache could not be found in env!", Seems to think I do not have memcache setup -- however I do.  Full error is here
*** Sander^work has quit IRC21:13
sniperdsearch only returns setting "allow_account_management = true" in config but its already set21:14
*** Sander^work has joined #openstack21:14
*** rnorwood has joined #openstack21:14
sniperdswift 1.2.0 installed from PPA ubuntu packages21:14
*** sniperd has quit IRC21:15
*** sniperd has joined #openstack21:15
uvirtbotNew bug: #899396 in tempest "KeyError masks real errors in server tests" [Medium,In progress]
*** jeremyb has quit IRC21:18
*** doude has quit IRC21:18
*** jeremyb has joined #openstack21:18
vicelowwich filesystem is recommended to use in a object server?21:19
*** doude has joined #openstack21:19
*** TheOsprey has quit IRC21:20
*** sniperd has quit IRC21:21
*** sniperd has joined #openstack21:21
zykes-vicelow: xfs21:22
vicelowthanks zykes-21:22
*** nati2 has joined #openstack21:22
*** lorin1 has joined #openstack21:23
koolhead17zykes-: sir21:26
koolhead17annegentle: around21:26
*** gnu111 has quit IRC21:26
notmynamesniperd: 2 things: 1) the cache error is probably because you don't have the cache middleware in the pipeline 2) 1.2.0 is pretty old. you should use a newer version. talk to mtaylor or ttx about the right PPA to pull it from21:27
zykes-koolhead17: ?:p21:27
mtaylorsniperd: what ppa are you installing from?21:28
sniperdnotmyname: Ah... OK21:28
koolhead17zykes-: how are you sirrrrrrrr21:28
koolhead17hola mtaylor21:28
mtaylorhola koolhead1721:28
koolhead17mtaylor: am good. how have you been?21:28
mtaylorsniperd: try using ppa:openstack-release/2011.3 instead21:28
mtaylorkoolhead17: excellent21:29
koolhead17mtaylor: sounds great!! :)21:29
sniperdmtaylor: was using
zykes-koolhead17: good21:30
koolhead17zykes-: nice21:30
mtaylorsniperd: that is very deprecated21:30
mtaylornotmyname: do you have an opposition to me deleting that ppa?21:30
sniperdmtaylor: that probably explains my problem(s) then, Thanks for the help, I appreciate it21:31
mtaylorsniperd: there is also a release ppa and a trunk ppa off of swift-core ... did you follow documentation recently which told you to use the one you're using? I'd like to fix it if possible21:31
annegentlekoolhead17: hi, am now!21:31
koolhead17annegentle: awesome21:32
notmynamemtaylor: I see no reason to keep the deprecated ppa around. I seriously doubt anyone is using it for prod (but I suppose it's possible), and I don't want them to use 1.2 anyway.21:32
*** yamahata has quit IRC21:33
notmynamesniperd: and to echo mtaylor, please let us know where you saw a reference to that old PPA. we need to make sure the docs are updated21:33
sniperdmtaylor: It's actually from a chef cookbook I found from voxel, Figured I would use that as a starting spot.
*** yamahata has joined #openstack21:34
sniperdIf I get this in good working order I will submit it to opscode so theres a semi-official one21:35
mtaylorsniperd: is there no swift coverage in openstack-chef?21:36
mtaylornope. great21:36
mtaylorsniperd: if you get it working, would you submit it to openstack/openstack-chef/21:37
sniperdmtaylor: yeah - I definitely will21:37
mtaylorsniperd: --- code submission guidelines at
*** dysinger has joined #openstack21:38
sniperdperfect, thanks again for the help21:38
mtaylornotmyname: done21:38
notmynamemtaylor: thanks21:38
*** debo-os has joined #openstack21:40
*** nerens_ has joined #openstack21:41
*** nerens has quit IRC21:41
*** nerens_ is now known as nerens21:41
*** ziyadb has quit IRC21:43
*** ziyadb has joined #openstack21:44
*** joesavak has joined #openstack21:45
*** Vinsh_HP has quit IRC21:46
*** mdomsch has joined #openstack21:51
uvirtbotNew bug: #899408 in swift "whitespace and line length violations" [Undecided,New]
*** lorin1 has quit IRC21:57
*** joesavak has quit IRC21:57
*** jfluhmann has joined #openstack21:57
*** codepoet has quit IRC21:58
*** mgoldmann has quit IRC22:00
*** adam_g has joined #openstack22:00
*** lts has quit IRC22:01
*** lmh has quit IRC22:03
*** lmh has joined #openstack22:04
*** llang629 has joined #openstack22:05
*** bencherian has quit IRC22:05
*** rsampaio has quit IRC22:06
*** bencherian has joined #openstack22:07
*** reed has quit IRC22:11
*** Guest3171 has quit IRC22:11
*** dolphm has quit IRC22:12
*** llang629 has left #openstack22:14
juddmsniperd: as the author of those voxel recipes, I'd like to encourge you to look at andi abes's chef recipes instead. :)22:15
*** dragondm has quit IRC22:15
juddmIt's rolled into crowbar now, but they work pretty OK without crowbar.22:15
*** vicelow has quit IRC22:15
adam_gheya, is keystone.conf's default_store unreleated to the database configured in sql_connection?22:16
*** juddm has quit IRC22:16
*** rsampaio has joined #openstack22:18
*** russellb has quit IRC22:19
*** russellb has joined #openstack22:19
*** reed has joined #openstack22:22
*** dolphm has joined #openstack22:25
uvirtbotNew bug: #899414 in nova "whitespace and line length violations" [Undecided,New]
*** dubsquared has quit IRC22:26
maplebedI'm trying to throw out some log messages from within one of the wsgi modules in swift-proxy.  can anybody offer advice on how to do that?22:28
*** vladimir3p has quit IRC22:28
notmynamemaplebed: if you are using the standard get_logger call in swift, anything you print to stdout will get logged (INFO level)22:29
*** vernhart has quit IRC22:29
maplebedthat's in swift.common.utils!  let's see if I can get that to work for me.22:30
*** Vinsh_HP has joined #openstack22:32
*** dolphm has quit IRC22:32
*** bcwaldon has quit IRC22:38
*** miclorb_ has joined #openstack22:39
maplebedoffhand, notmyname, do you know if I need to kick the swift-proxy on each change to my python or does it pick up changes live?22:39
notmynamemaplebed: you'll need to reload the proxy so it sees the updated middleware. swift-init proxy reload will do a graceful close with no client interruption. swift-init proxy restart will terminate existing connections and restart the process22:40
* AaronSchulz looks up CDN sharing22:43
notmynamemaplebed: obviously, use a different name other than "proxy" if you are doing middleware on the other servers...22:44
*** lborda has quit IRC22:44
*** vernhart has joined #openstack22:44
notmynameAaronSchulz: what is CDN sharing?22:45
notmynameAaronSchulz: heh. that should be there22:46
notmynameannegentle: ^^ looks like there is a holdover from the cloud files docs22:46
sniperdjuddm: will do :)22:47
maplebednotmyname: you mean "that shouldn't be there", right?22:48
notmynameAaronSchulz: that's also old docs22:48
notmynamemaplebed: oh yeah22:48
notmynameAaronSchulz: heh. that shouldn't be there22:48
AaronSchulzI saw make_private() and is_public in cloudfiles.php when I was looking for access controls22:48
*** ryker has quit IRC22:48
notmynameAaronSchulz: the cdn stuff (in those old docs) is a holdover from where those docs came from (the rackspace cloud files dev guide). swift doesn't include any cdn integration22:49
notmynameAaronSchulz: it seems the current docs are correct, though22:49
AaronSchulzis there a way to restrict access of containers to certain users (without custom python code)22:49
notmynameAaronSchulz: that's up to the auth server you use.22:50
notmynameAaronSchulz: example for tempauth (and swauth):
*** lloydde has joined #openstack22:52
*** kbringard has quit IRC22:56
AaronSchulzmaplebed: I'm thinking that the auth for rewritten old-style urls should use a different user than when MW writes/reads things from the proxy via cloudfiles22:56
AaronSchulznotmyname: thanks22:57
maplebedsure.  that'd let us give read-only vs. read-write perms to the files.22:58
AaronSchulzcontainers like xx-wikipedia-images-deleted would be restricted22:58
*** bencherian has quit IRC23:02
*** bencherian has joined #openstack23:08
*** jj0hns0n has quit IRC23:08
*** GheRivero has quit IRC23:08
*** gyee has quit IRC23:11
*** rustam has joined #openstack23:12
*** ahasenack has quit IRC23:16
*** RicardoSSP has joined #openstack23:17
*** MarkAtwood has quit IRC23:17
*** debo-os_ has joined #openstack23:17
*** miclorb_ has quit IRC23:18
*** rnorwood has quit IRC23:19
*** debo-os has quit IRC23:20
*** debo-os_ is now known as debo-os23:21
*** jakedahn has joined #openstack23:21
*** jdurgin has quit IRC23:26
*** debo-os has quit IRC23:27
*** jdurgin has joined #openstack23:28
*** dolphm has joined #openstack23:28
*** imsplitbit has quit IRC23:41
*** jfluhmann has quit IRC23:44
*** code_franco has quit IRC23:47
*** koolhead17 has left #openstack23:47
*** dolphm has quit IRC23:50
*** rnirmal has quit IRC23:55
*** ziyadb has quit IRC23:56
*** deshantm_laptop_ has joined #openstack23:58

Generated by 2.14.0 by Marius Gedminas - find it at!