MySqlNode ... IllegalStateException ... Text file busy.

[David Toy]

Folks,

While working through the YAML getting started instructions I hit this problem. (Once, of three runs).

I've not seen this before, and wondered if anyone else could shed some light?

https://gist.github.com/jdtoy/10723327

Error invoking start at MySqlNodeImpl{id=kJZb7nfB}: org.jclouds.compute.RunNodesException: error running 1 node group(brooklyn-mjzn-davidtoy-my-web-cluste-yqth-mysql-kjzb) location(us-east-1) image(ami-7f948916) size(m1.small) options({inboundPorts=[22, 3306], scriptPresent=true, userMetadata={Name=brooklyn-mjzn-davidtoy-my-web-cluste-yqth-mysql-kjzb-Dq6a}, userDataCksum=2f4a740b})
Execution failures:

 
0 error[s]

Node failures:
 

1) IllegalStateException on node us-east-1/i-15958144:

java.lang.IllegalStateException: error running [/tmp/init-bootstrap init] as ubu...@54.81.194.216; returnVal !=0: {output=bash: /tmp/init-bootstrap: /bin/bash: bad interpreter: Text file busy
, error=, exitStatus=126}



Thanks,
David

[Alex Heneveld]

// cross-posting to dev@jclouds and brooklyn-dev as I think this is a weird jclouds edge case

In Brooklyn we're seeing what looks like a funky jclouds VM setup issue -- see original mail from David Toy below.  The error is:

1) IllegalStateException on node us-east-1/i-15958144:
java.lang.IllegalStateException: error running [/tmp/init-bootstrap init] as ubu...@54.81.194.216; returnVal !=0: {output=bash: /tmp/init-bootstrap: /bin/bash: bad interpreter: Text file busy
, error=, exitStatus=126}

Based on [1] and [2] it looks like init-bootstrap.sh is not always closed before it is executed.  This is an AWS machine but presumably there is nothing AWS specific about it.

I'm guessing it's due to a rogue slow write on the cloud machine and worth opening a jclouds issue for this -- and looking for a fix!  But before doing that I wanted to sanity check and gauge urgency.  So far it has only happened once.  Has anyone else seen this?

Thanks
Alex

[1] http://stackoverflow.com/questions/1384398/usr-bin-perl-bad-interpreter-text-file-busy
[2] https://dpk.net/2009/03/13/flock-before-execve/

--
You received this message because you are subscribed to the Google Groups "brooklyn-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to brooklyn-dev...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Alex Heneveld]

Ignasi- Almost certainly sshj. Thanks for confirming that this is as weird as we thought. And likely hard to test / reproduce.

David- anything unusual in the deployment? is it an m1.micro?

Thinking maybe the shell closing the file is slow eg cloud disk contended. So we need to look at jclouds mechanism for closing the file. The thing is we do lots of script copy and run type ops and have never seen this.

In any case let's open an issue.

Best
Alex

On 15 Apr 2014 09:25, "Ignasi Barrera" <na...@apache.org> wrote:

I've never seen this and seems like a race condition. A quick Google search
shows that the error appears when you try to execute a script file that is
open for writing.

I'm wondering why the file is still open when the script is executed and it
seems that you're right Alex.

Out of curiosity, which SSH driver is being used?
El 15/04/2014 17:12, "Alex Heneveld" <alex.h...@cloudsoftcorp.com>
escribió:

>> init] as ubu...@54.81.194.216 <mailto:ubu...@54.81.194.216>; returnVal

>> !=0: {output=bash: /tmp/init-bootstrap: /bin/bash: bad interpreter: Text
>> file busy
>>
>> , error=, exitStatus=126}
>> Thanks,
>> David
>> --
>> You received this message because you are subscribed to the Google Groups
>> "brooklyn-dev" group.
>> To unsubscribe from this group and stop receiving emails from it, send an

>> email to brooklyn-dev...@googlegroups.com <mailto:brooklyn-dev+
>> unsub...@googlegroups.com>.