On 09/18/2012 12:44 AM, Prakash Ranjan wrote:
Hi All,
After deploying any image on my local lava-server setup, in test
image , the resolv.conf file is always missing.
Here is the complete log link. http//paste.ubuntu.com/1212314/
You can check the job fiel at this link.. http//paste.ubuntu.com/1212328
My master image info : http//paste.ubuntu.com/1212330.
Please help me on this issue.
This is a bit of a guess, but I noticed in your job definition you:
1. flash image 2. install lava-test 3. boot 4. run lava-test
Our jobs normally do those 4 steps but in this order:
1. flash image 2. boot 3. install lava-test 4. run lava-test
ie steps 2 and 3 are reversed. My guess is that the "boot" step is creating this file so things work for us. You could probably just try that. However, when I look at a current linaro RFS like today's nano image, I see a /etc/resolv.conf in it.
We could probably also patch the dispatcher so it doesn't fail if the file doesn't exist.
-andy
On 18 Sep 2012, at 19:44, Andy Doan wrote:
On 09/18/2012 12:44 AM, Prakash Ranjan wrote:
Hi All,
After deploying any image on my local lava-server setup, in test
image , the resolv.conf file is always missing.
Here is the complete log link. http//paste.ubuntu.com/1212314/
You can check the job fiel at this link.. http//paste.ubuntu.com/1212328
My master image info : http//paste.ubuntu.com/1212330.
Please help me on this issue.
This is a bit of a guess, but I noticed in your job definition you:
- flash image
- install lava-test
- boot
- run lava-test
Our jobs normally do those 4 steps but in this order:
- flash image
- boot
- install lava-test
- run lava-test
ie steps 2 and 3 are reversed. My guess is that the "boot" step is creating this file so things work for us. You could probably just try that. However, when I look at a current linaro RFS like today's nano image, I see a /etc/resolv.conf in it.
We could probably also patch the dispatcher so it doesn't fail if the file doesn't exist.
Yeah, I pointed the order issue out to prak on irc yesterday, but he pointed out that there are jobs which do it the other way and succeed, so there's something else going on, and I suspect that it's something to do with the way prak is cacheing the hwpacks and rootfs on his system, but it's difficult to diagnose at a distance.
Prak filed a bug yesterday to that effect, i.e. to ignore failure on copy, but that feels like a lash up fix because we haven't fully understood what's going on.
I'll talk to prak again on irc today and see if we can't nail it.
Thanks
Dave
On 09/19/2012 02:10 AM, Dave Pigott wrote:
On 18 Sep 2012, at 19:44, Andy Doan wrote:
On 09/18/2012 12:44 AM, Prakash Ranjan wrote:
Hi All,
After deploying any image on my local lava-server setup, in test
image , the resolv.conf file is always missing.
Here is the complete log link. http//paste.ubuntu.com/1212314/
You can check the job fiel at this link.. http//paste.ubuntu.com/1212328
My master image info : http//paste.ubuntu.com/1212330.
Please help me on this issue.
This is a bit of a guess, but I noticed in your job definition you:
- flash image
- install lava-test
- boot
- run lava-test
Our jobs normally do those 4 steps but in this order:
- flash image
- boot
- install lava-test
- run lava-test
ie steps 2 and 3 are reversed. My guess is that the "boot" step is creating this file so things work for us. You could probably just try that. However, when I look at a current linaro RFS like today's nano image, I see a /etc/resolv.conf in it.
We could probably also patch the dispatcher so it doesn't fail if the file doesn't exist.
Yeah, I pointed the order issue out to prak on irc yesterday, but he pointed out that there are jobs which do it the other way and succeed, so there's something else going on, and I suspect that it's something to do with the way prak is cacheing the hwpacks and rootfs on his system, but it's difficult to diagnose at a distance.
Prak filed a bug yesterday to that effect, i.e. to ignore failure on copy, but that feels like a lash up fix because we haven't fully understood what's going on.
I'll talk to prak again on irc today and see if we can't nail it.
I added a comment to his bug. Antonio spotted this same thing in our lab yesterday. So its not an isolated bug. Should be pretty easy to fix though.
linaro-validation@lists.linaro.org