Could not find data source

[raymdt]

Hi and thanks your efforts.

but I am able to successfully run the Maverick image on my NC outside of the cloud

kvm without VT just runs QEMU. Please check the VT in Bios and run kvm-ok on NC.

Sometime it is easier to reinstall eucalyptus(it takes just 2 hours).

[nbartolome]

Hi and thanks your efforts.



kvm without VT just runs QEMU. Please check the VT in Bios and run kvm-ok on NC.

Sometime it is easier to reinstall eucalyptus(it takes just 2 hours).

I double checked the bios and verified that Virtualization is enabled.

kvm-ok looks okay.

cloud@uec02:/etc/eucalyptus$ sudo kvm-ok
INFO: Your CPU supports KVM extensions
INFO: /dev/kvm exists
KVM acceleration can be used

I have re-installed multiple times. I'm actually in the process of re-installing again.

[nbartolome]

Okay, so I got my UEC on Ubuntu 10.10 to work. In my previous setup my CLC and NC were physically connected to a Cisco 3548 with multiple VLANs and a somewhat "secure" setup in that it requires us to set up static routes in order for all the other VLANs to pass traffic to one another. Even though I setup static routes for 172.19.0.0 and 169.254.169.254 to be able to communicate to the CLC and NC i was still receiving this error.

So I moved the CLC and NC to a "dumb" switch without any VLANs (similar to a cheesy Netgear 5 port switch) and re-installed. The result is that I have a working cloud and can successfully log in to instances, run apache webserver on them and access web pages from my desktop.

Here are a few interesting things that I've discovered:

• Even though UEC assigns a 172.19.1.2, 172.19.1.3, etc.. address you will not ever see it on the CLC or the NC. The only place that I have found it is on the actual instance. It looks like the NC i brings up a 'vnetX" interface and you can confirm the mac address with the one being used in your instance.
• I am not administering my cloud via the command line on the CLC. I am using the hybridfox plugin for firefox.
• I am not launching my instances as the default 'admin' user. Instead I created a different user through the web interface and gave it Administrator privileges.
• Although this new user has Admin privileges I cannot download images from the store because I get a "bad request signature" error. So in order to download an image from the store I have to login as the "admin" user.
• The only image I can get successfully to run is Lucid (amd64). Both Karmic (AMD 64) from the store nor Maverick downloaded from uec-images.ubuntu.com will run. They just stay in "pending" state until they terminate themselves.
• I sometimes receive this error on the instance console: "consuming user data failed!" Apparently this is a bug with the image that has been fixed but hasn't been pushed out to the store yet ? http://open.eucalyptus.com/forum/possible-bug-walrus-20 . Suggestions say to just keep trying to launch and it'll eventually run successfully. I have confirmed this and after the 4th or 5th time will finally get the instance to boot up successfully.

I bet I could have gotten it to work on my Cisco switch. The switch ports probably just need to be set to Operational mode multi or something like that.

I hope this can help any other people that have been working through the bugs.

[kim0]

I believe UEC requires a vlan clean switch. I'm copying this from a whitepaper

"you will also need to verify that yournetwork is “VLAN-clean”, that is, it passes Ethernet frame with VLAN tag intact. Managed switches will often be configured to strip the VLAN tag on frame it pass through, or to drop these frames entirely. As security groups is implemented in Eucalyptus using special VLAN interface,you will need to configure your switches to pass Ethernet frames between the Node Controllerand the front-end as-is."

[nbartolome]

Yeah that's what made me suspect the network setup because when I went to test VLAN clean my machines would hang and become unresponsive.

http://open.eucalyptus.com/wiki/Euca...iguration_v2.0

However I thought it was only for MANAGED mode and by default it installs as MANAGED-NOVLAN. I also tried to use STATIC and SYSTEM but those were also unsuccessful.

Note that in MANAGED mode the underlying physical network must be VLAN clean, as Eucalyptus provides and manages its own VLAN tagging. If your network is not VLAN clean, you can use MANAGED-NOVLAN mode, which provides the full set of networking features, with the exception of VM isolation between instances.
In the remaining networking modes, SYSTEM and STATIC, there are no virtual subnets—VM instances appear on the physical network as if they were physical machines; and VM instances are directly bridged with the NC machine's physical ethernet device.

[rabinnh]

I am having the identical problem, but I am running in SYSTEM mode. One image instance hangs on "Waiting for EC2 meta-data" and the other gives me this:

Begin: Running /scripts/init-bottom ...
Done.
[ 3.984801] EXT3 FS on sda1, internal journal
sleeping 1
sleeping 1
sleeping 1
sleeping 1
sleeping 1
sleeping 1
sleeping 1
sleeping 1
sleeping 1
sleeping 1
Could not find data source
Failed to get instance datainit: cloud-init main process (620) terminated with status 1
mountall: Event failed
* Starting AppArmor profiles Traceback (most recent call last):
File "/usr/bin/cloud-init-cfg", line 56, in <module>
main()
File "/usr/bin/cloud-init-cfg", line 43, in main
cc = cloudinit.CloudConfig.CloudConfig(cfg_path)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 42, in __init__
Traceback (most recent call last):
File "/usr/bin/cloud-init-cfg", line 56, in <module>
main()
File "/usr/bin/cloud-init-cfg", line 43, in main
cc = cloudinit.CloudConfig.CloudConfig(cfg_path)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 42, in __init__
self.cfg = self.get_config_obj(cfgfile)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 53, in get_config_obj
f=file(cfgfile)
IOError: [Errno 2] No such file or directory: '/var/lib/cloud/data/cloud-config.txt'
self.cfg = self.get_config_obj(cfgfile)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 53, in get_config_obj
f=file(cfgfile)
IOError: [Errno 2] No such file or directory: '/var/lib/cloud/data/cloud-config.txt'
Traceback (most recent call last):
File "/usr/bin/cloud-init-cfg", line 56, in <module>
main()
File "/usr/bin/cloud-init-cfg", line 43, in main
cc = cloudinit.CloudConfig.CloudConfig(cfg_path)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 42, in __init__
self.cfg = self.get_config_obj(cfgfile)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 53, in get_config_obj
f=file(cfgfile)
IOError: [Errno 2] No such file or directory: '/var/lib/cloud/data/cloud-config.txt'
[ OK ]
Traceback (most recent call last):
File "/usr/bin/cloud-init-cfg", line 56, in <module>
main()
File "/usr/bin/cloud-init-cfg", line 43, in main
cc = cloudinit.CloudConfig.CloudConfig(cfg_path)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 42, in __init__
self.cfg = self.get_config_obj(cfgfile)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 53, in get_config_obj
f=file(cfgfile)
IOError: [Errno 2] No such file or directory: '/var/lib/cloud/data/cloud-config.txt'
Traceback (most recent call last):
File "/usr/bin/cloud-init-cfg", line 56, in <module>
main()
File "/usr/bin/cloud-init-cfg", line 43, in main
cc = cloudinit.CloudConfig.CloudConfig(cfg_path)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 42, in __init__
self.cfg = self.get_config_obj(cfgfile)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 53, in get_config_obj
f=file(cfgfile)
IOError: [Errno 2] No such file or directory: '/var/lib/cloud/data/cloud-config.txt'
Traceback (most recent call last):
File "/usr/bin/cloud-init-cfg", line 56, in <module>
main()
File "/usr/bin/cloud-init-cfg", line 43, in main
cc = cloudinit.CloudConfig.CloudConfig(cfg_path)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 42, in __init__
self.cfg = self.get_config_obj(cfgfile)
File "/usr/lib/python2.6/dist-packages/cloudinit/CloudConfig.py", line 53, in get_config_obj
f=file(cfgfile)
IOError: [Errno 2] No such file or directory: '/var/lib/cloud/data/cloud-config.txt'


Since the errors are different, is it really probable that it is a switch stripping vlan tags? Especially since I am running in SYSTEM mode.

[kim0]

My advice would be not to run in SYSTEM mode, since that does not get enough testing. Keep the default networking mode

[rabinnh]

I started in MANAGED-NOVLAN mode and all the instances stayed in PENDING. I can't really understand why there is documentation for things that (at least you are saying) don't work.

I have seen this reply on these forums quite a bit, so I'd like some clarification. Do you work for Canonical? I'm just curious how you know about the test plan and test matrix.

Thanks

[kim0]

Hi, I'm not saying it doesn't work, I'm just saying other networking modes are not getting enough testing, thus are most likely to be buggy. The advice is to stick to the default networking mode. Yes I work for canonical, although that's irrelevant. I got this information from the Ubuntu server team on irc on #ubuntu-cloud you could join there and will probably get the same info

[rabinnh]

I went back to 10.04 and Eucalyptus 1.6.2 and had everything up and working in about an hour. I spent 4 days trying to get 10.10 to work on the same hardware.

I think some (but not all) of the problems are caused by the images in the Store. If they are to remain in the Store for 10.10 then someone should ensure that they work properly. I don't think that people are going to attempt to bundle their own images when they can't get images that are blessed by Ubuntu to work based on the posts that I have been reviewing.

I see that Ubuntu is going to support OpenStack in 11.04. I say, "good move". Eucalyptus is not up to your standards in terms of documentation, robustness, and testing. I don't think that UEC can be fixed by Canonical because the quality of Eucalyptus is questionable. It's a great idea, but it doesn't seem to have gone through a professional level of review, testing, and debugging before release.

Regards