Re: skip email on transient build failures

> Date: Fri, 13 Aug 2010 12:01:39 -0400
> From: Dan Rollo <[email protected]>
> 
> Hi David,
> 
> Interesting idea. Unfortunately, I'm not sure how you would persist such 
> state as the # of failures across email publishers (because each build 
> creates new objects).

Since CC knows how to not spam on repeat failures, then email
on success, the information must sit somewhere, right?

> I'm sure there's a way, but it doesn't jump out at 
> me...might need some new way to do it.
> 
> If you take a crack at writing a patch to provide this feature, please 
> be sure to create a Jira issue with your patch attached (and of course 
> holler if you get stuck).

Actually, we do have some folks who have dug into CC and patched a few
things, so Maybe I can persuade them to give it a try.

> Another thought: would it be easier to built some kind of retry into the 
> steps with the intermittent failures? eg: retry the repo fetch?

i.e. last night, we had a failure due to the following:

    surefire:test {execution: default-test}
    
    Error occurred during initialization of VM
    Could not reserve enough space for object heap

but the rebuild ran just fine a couple minutes later. Completely unrelated to
Maven repo timeouts. So I'd really have to just build the retry into <maven2> 
to catch all possible problems... but that does not help folks who have
similar problems but don't use <maven2>. I'd rather just rely on the normal CC 
retry (including creating logs showing Failed) and just confine the affect
to <email> (and by extension <htmlemail>).

thanks

> Dan
> 
> Date: Thu, 12 Aug 2010 10:02:04 -0400
> Subject: [Cruisecontrol-user] skip email on transient build failures
> To: [email protected]
> 
> 
> It's not uncommon for our CC builds to fail for transient reasons
> (usually it is timeouts from Maven central repository, despite our use 
> of Artifactory). [...]
> 
> I propose an enhancement to the <email> publisher to only send mail
> on the second consecutive failure. That is, do not email on the first
> failure and if the next build succeeds, do not send a success email.
> Only send mail if a rebuild after failure also fails.
> [...]
>    <ignoreFirst>emailaddress</ignorefirst> (cardinality 0 .. *)

-- 
David J. Biesack, SAS
SAS Campus Dr. Cary, NC 27513
www.sas.com    (919) 531-7771

------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev