To Chef 11 or not to Chef 11


#1

A discussion surrounding Chef 10 support has surfaced in the knife-server
project. If you’re not aware, knife-server is Fletcher Nichol’s project to
provide a knife plugin to build chef servers similar to how knife bootstrap
works.

While the comment is against my fork, it’s surrounding some patches that
him, Dan Ryan and myself are working on that both add Chef 11 support and
some other platform independence features.

https://github.com/erikh/knife-server/pull/3#issuecomment-13363152

The main thing to take away from this is that the chef-server bootstrap may
or may not require chef 11 at some point in the near future.

I guess what concerns me is the official state of Chef 10 support in the
wake of the Chef 11 release. I don’t think anyone’s claiming Seth’s
comments are official or even necessarily authoritative, but if this is to
be relied on I think there are some concerns here that need to be
addressed, namely, what the status of those relying on Chef 10 can expect.

To put this in context, as late as last April I was working with a team to
get off of Chef 0.9, and that isn’t a unique situation by any sense of the
word. I don’t think reasoning about the EOL of Chef 10, or even its
supported nature, two weeks after Chef 11’s GA release is something anyone
should have to be doing.

The ideal thing I’d like to get out of this is some kind of minimal effort
to ensure those who need Chef 10 can still get it installed easily.

I think the easiest thing would just be to allow us to point at
bootstrap-$version.tar.gz for things that require a specific major version
of chef. Theoretically, these should be pretty stable over time so the
amount of maintenance burden should be pretty low for you guys as the
distance between $version and the current state of things grows. And of
course when things are EOL’d what ever happens, happens.

There might be a better solution that revolves around omnibus too, but
having that cross-platform bit is one of the reasons the chef-solo
bootstrap is so attractive. To be clear though I don’t really care too much
as long as I can still install Chef 10 on something that doesn’t dictate my
OS too. :slight_smile:

Anyhow, I’d love to hear if there’s any official stance on this or what
people at opscode think so we can make more accurate decisions on how to
support Chef 10 for future releases of knife-server.

Thanks!

-Erik


#2

On Monday, February 11, 2013 at 3:37 AM, Erik Hollensbe wrote:

A discussion surrounding Chef 10 support has surfaced in the knife-server project. If you’re not aware, knife-server is Fletcher Nichol’s project to provide a knife plugin to build chef servers similar to how knife bootstrap works.

While the comment is against my fork, it’s surrounding some patches that him, Dan Ryan and myself are working on that both add Chef 11 support and some other platform independence features.

https://github.com/erikh/knife-server/pull/3#issuecomment-13363152

The main thing to take away from this is that the chef-server bootstrap may or may not require chef 11 at some point in the near future.

I guess what concerns me is the official state of Chef 10 support in the wake of the Chef 11 release. I don’t think anyone’s claiming Seth’s comments are official or even necessarily authoritative, but if this is to be relied on I think there are some concerns here that need to be addressed, namely, what the status of those relying on Chef 10 can expect.

To put this in context, as late as last April I was working with a team to get off of Chef 0.9, and that isn’t a unique situation by any sense of the word. I don’t think reasoning about the EOL of Chef 10, or even its supported nature, two weeks after Chef 11’s GA release is something anyone should have to be doing.
For Chef 0.9, we initially planned support for 6 months but stretched that out to a year.

We’ll make an announcement about Chef 10 lifecycle soon. Currently the plan is to support Chef 10 for a year, but this decision isn’t final yet.


Daniel DeLeo


#3

Alright – the time isn’t really as important as how we’ll be able to
reason about installing chef 10 servers in the future – is this something
we can sort out independently?

I totally get the maintenance woes and the need to EOL; it’s more about how
I can expect to get Chef 10 server installed in a reliable way (and for
these open source projects, in a way that we can ship to others).

On Mon, Feb 11, 2013 at 7:41 AM, Daniel DeLeo dan@kallistec.com wrote:

On Monday, February 11, 2013 at 3:37 AM, Erik Hollensbe wrote:

A discussion surrounding Chef 10 support has surfaced in the knife-server
project. If you’re not aware, knife-server is Fletcher Nichol’s project to
provide a knife plugin to build chef servers similar to how knife bootstrap
works.

While the comment is against my fork, it’s surrounding some patches that
him, Dan Ryan and myself are working on that both add Chef 11 support and
some other platform independence features.

https://github.com/erikh/knife-server/pull/3#issuecomment-13363152

The main thing to take away from this is that the chef-server bootstrap
may or may not require chef 11 at some point in the near future.

I guess what concerns me is the official state of Chef 10 support in the
wake of the Chef 11 release. I don’t think anyone’s claiming Seth’s
comments are official or even necessarily authoritative, but if this is to
be relied on I think there are some concerns here that need to be
addressed, namely, what the status of those relying on Chef 10 can expect.

To put this in context, as late as last April I was working with a team to
get off of Chef 0.9, and that isn’t a unique situation by any sense of the
word. I don’t think reasoning about the EOL of Chef 10, or even its
supported nature, two weeks after Chef 11’s GA release is something anyone
should have to be doing.

For Chef 0.9, we initially planned support for 6 months but stretched that
out to a year.

We’ll make an announcement about Chef 10 lifecycle soon. Currently the
plan is to support Chef 10 for a year, but this decision isn’t final yet.


Daniel DeLeo