Re: [Fab-user] ssh.transport

[Samuli Seppänen]

Il 17.03.2012 01:56, Maximilia Black ha scritto:

Hi,

I'm having an error when I try to connect to one of my servers through fabric.  I keep getting the following error:

No handlers could be found for logger "ssh.transport"

Fatal error: Error reading SSH protocol banner

Aborting

I'm hoping someone has come across something like this that can shed some light?  I need to instantiate dozens of ec2

instances at once and through fabric update some applications.  Now on some of the instances everything works fine but on others I get this error, and it's at random.  When I retry to do the same command (like run('something')) after a minute it works okay...

Any help on this would be greatly appreciated.

MB

Hi,

I think the EC2 VMs have not initialized properly when you're trying to run the command.

I've used Fabric to configure up to 100 newly created EC2 VM. To make sure all EC2 VMs are up before running Fabric, I wrote a piece of Python+Boto code which

- uses Boto to launch the n number of EC2 VMs
- gets a (filtered) list of VMs and check if all are up
- if not, waits 10 seconds and repeats the check

Once it sees all VMs are up, it launches Fabric. That way all EC2 VMs are guaranteed to be up before running any commands. If you're interested in this tool, look here:

<https://community.openvpn.net/openvpn/wiki/PerformanceTesting#Tools>

I'd need to upgrade it to use Fabric's parallel mode (as well as EC2 tags) to make it simpler and cleaner. Atm it handles parallelism itself, and is thus somewhat complex.

Samuli