Suggestions for upcoming community summit


#1

Unfortunately, I will not be able to attend the community summit. I imagine
there are some others in the same boat as myself.

It would be great if the event could be livestreamed for this reason.

I have written up just a few suggestions for the event, such as live
streaming and topics to address.

It would be great if others could add their ideas well
http://wiki.opscode.com/display/chef/Suggestions+for+Community+Summit

I have just a few items. Here is what is there so far

  • How can we make the cookbook
    sitehttp://community.opscode.com/cookbooks and
    cookbook community as useful and vibrant as that of rubygems?
  • What are anti-patterns for cookbooks?
  • What are best practices for cookbooks?
  • Performance Tuning
  • Building on top of chef, a la crowbar
  • Cookbook testing and how to automate it, esp. for opscode/cookbooks
  • What should the role of opscode/cookbooks be?
  • Professional certification for Chef, like the MCSE, RHCE. Should Chef
    have it? How can we make it not suck

#2

Some great ideas. As a new user just developing my use of the
platform, I’d like to throw out some things from what may be a
different perspective:

  1. Clustering of the server. How to make it resilient in the face of
    cloud providers SLA levels.
  2. Managing clusters. I’m taken with the cluster_chef recipes that
    infochimps is developing. Defining infrastructure as clusters that
    are independent but which can communicate seems a natural extension of
    infrastructure as code.
  3. Making configuration more transparent. It’d be nice to have a tool
    that could spit out graphviz or something similar to display
    dependencies and actions on a system.
  4. Auditing/intentions of configuration changes are opaque. There is
    a rich set of data that chef sits in the middle of, but it doesn’t
    seem like it does a lot of reporting on this data. Izf I want to know
    who uploaded a recipe and when it was applied to a node, I’d like to
    have that come from “knife audit cookbook” and “knife audit node” and
    see a history of what changes have been made.
  5. Sizing and scaling a chef server/infrastructure. Redundancy, CYA, etc.
  6. How to best support opscode, chef, the community. What options are
    there to support opscode and chef to ensure a thriving community?
    Sure, there’s “use the service” but that’s not always the best
    business decision. What other ways can we help support the devs?

-Peter

On Tue, Nov 1, 2011 at 7:14 AM, Bryan Berry bryan.berry@gmail.com wrote:

Unfortunately, I will not be able to attend the community summit. I imagine
there are some others in the same boat as myself.
It would be great if the event could be livestreamed for this reason.
I have written up just a few suggestions for the event, such as live
streaming and topics to address.

It would be great if others could add their ideas well
http://wiki.opscode.com/display/chef/Suggestions+for+Community+Summit
I have just a few items. Here is what is there so far

How can we make the cookbook site and cookbook community as useful and
vibrant as that of rubygems?
What are anti-patterns for cookbooks?
What are best practices for cookbooks?
Performance Tuning
Building on top of chef, a la crowbar
Cookbook testing and how to automate it, esp. for opscode/cookbooks
What should the role of opscode/cookbooks be?
Professional certification for Chef, like the MCSE, RHCE. Should Chef have
it? How can we make it not suck


#3

Sure, there’s “use the service” but that’s not always the best
business decision. What other ways can we help support the devs?

+1 on that. My organization doesn’t intend to purchase hosted chef but
there should be some way for such organizations to help “invest” in the
future development of Chef.

On Tue, Nov 1, 2011 at 5:41 PM, Peter Norton pn+chef-list@knewton.comwrote:

Some great ideas. As a new user just developing my use of the
platform, I’d like to throw out some things from what may be a
different perspective:

  1. Clustering of the server. How to make it resilient in the face of
    cloud providers SLA levels.
  2. Managing clusters. I’m taken with the cluster_chef recipes that
    infochimps is developing. Defining infrastructure as clusters that
    are independent but which can communicate seems a natural extension of
    infrastructure as code.
  3. Making configuration more transparent. It’d be nice to have a tool
    that could spit out graphviz or something similar to display
    dependencies and actions on a system.
  4. Auditing/intentions of configuration changes are opaque. There is
    a rich set of data that chef sits in the middle of, but it doesn’t
    seem like it does a lot of reporting on this data. Izf I want to know
    who uploaded a recipe and when it was applied to a node, I’d like to
    have that come from “knife audit cookbook” and “knife audit node” and
    see a history of what changes have been made.
  5. Sizing and scaling a chef server/infrastructure. Redundancy, CYA, etc.
  6. How to best support opscode, chef, the community. What options are
    there to support opscode and chef to ensure a thriving community?
    Sure, there’s “use the service” but that’s not always the best
    business decision. What other ways can we help support the devs?

-Peter

On Tue, Nov 1, 2011 at 7:14 AM, Bryan Berry bryan.berry@gmail.com wrote:

Unfortunately, I will not be able to attend the community summit. I
imagine
there are some others in the same boat as myself.
It would be great if the event could be livestreamed for this reason.
I have written up just a few suggestions for the event, such as live
streaming and topics to address.

It would be great if others could add their ideas well
http://wiki.opscode.com/display/chef/Suggestions+for+Community+Summit
I have just a few items. Here is what is there so far

How can we make the cookbook site and cookbook community as useful and
vibrant as that of rubygems?
What are anti-patterns for cookbooks?
What are best practices for cookbooks?
Performance Tuning
Building on top of chef, a la crowbar
Cookbook testing and how to automate it, esp. for opscode/cookbooks
What should the role of opscode/cookbooks be?
Professional certification for Chef, like the MCSE, RHCE. Should Chef
have
it? How can we make it not suck


#4

Hey everyone,

  1. Bryan Berry took action & created a page to capture topic/session ideas for the Community Summit. Check it out here: http://wiki.opscode.com/display/chef/Suggestions+for+Community+Summit

  2. Bryan & a number of others suggested livestreaming and/or doing some other kind of real-time remote participation in the Summit.

The tl;dr here is that real-time remote participation in Open Space is challenging to do in a way that doesn’t interfere with and dampen the dynamic and intimate flow of Open Space.

If you have never participated in an Open Space, this may sound strange. The Opscode Community Summit isn’t a conference. There are no speakers, panelists, or keynotes. There is no pre-set agenda and there is no “audience”. There are only fellow participants, a unique and incredible group of people gathered together in one place to work on something that matters to all of us.

There are at least two challenges to livestreaming & real-time remote participation:

  1. The space we create is a safe place for people to openly discuss complex and often confidential topics without worrying that someone will share it, an idea Tim O’Reilly calls the FrieNDA. Livestream ing or recording sessions by default has the potential to change this, and in most cases just doesn’t make sense to do as the facilitator. If someone decides to propose a streamed or recorded session, host a call in show, or anything else they are fully empowered to do so.

  2. The technology required for people to participate in real-time and at scale, which preserves the power and intensity of a face-to-face interaction and doesn’t create friction that distracts from the experience… isn’t readily available. Many people find they close their laptops/ipads/etc entirely and focus on being present. If people want to host questions on IRC or other means, they are fully empowered to do so.

We will provide a wiki for participants to document sessions, issues, actions, conversations, and conclusions. While we don’t know what exactly what will happen over the two days of the Summit, there are some things that are guaranteed to happen:
• The issues that are most important to people will get discussed & worked on
• The issues raised will be addressed by the participants best capable of getting something done about them.
• All of the most important ideas, recommendations, discussions, and next steps will be documented in a report which will be published to the entire community.
• Participants will be deeply engaged, connected, and energized.

Hope this helps,

-Jesse


Jesse Robbins | Cofounder & Chief Community Officer - Opscode, Inc.
e: jesse@opscode.com | p: +1-206-755-3739 | http://www.opscode.com


#5

On Tuesday, November 1, 2011 at 9:48 AM, Bryan Berry wrote:

Sure, there’s “use the service” but that’s not always the best
business decision. What other ways can we help support the devs?

+1 on that. My organization doesn’t intend to purchase hosted chef but there should be some way for such organizations to help “invest” in the future development of Chef.
We’d certainly love to have more people involved with testing our beta and RC (pre-)releases. There’s a 0.10.6 beta available now, soon to be an RC if no issues are reported.


Dan DeLeo


#6

what would be the best to protect my current system before testing it?

off the top of my head

  1. back up couchdb
  2. back up gems directory
  3. backup /{var,usr}/lib/chef
  4. stop chef services
  5. put new chef code in place of old, depending on whether using gems or
    rpm
  6. restart chef services

did I miss anything in this list?

On Wed, Nov 2, 2011 at 6:39 AM, Daniel DeLeo dan@kallistec.com wrote:

On Tuesday, November 1, 2011 at 9:48 AM, Bryan Berry wrote:

Sure, there’s “use the service” but that’s not always the best
business decision. What other ways can we help support the devs?

+1 on that. My organization doesn’t intend to purchase hosted chef but
there should be some way for such organizations to help “invest” in the
future development of Chef.
We’d certainly love to have more people involved with testing our beta and
RC (pre-)releases. There’s a 0.10.6 beta available now, soon to be an RC if
no issues are reported.


Dan DeLeo


#7

On Tuesday, November 1, 2011 at 10:56 PM, Bryan Berry wrote:

what would be the best to protect my current system before testing it?

off the top of my head
back up couchdb
back up gems directory
backup /{var,usr}/lib/chef
stop chef services
put new chef code in place of old, depending on whether using gems or rpm
restart chef services

did I miss anything in this list?
Backing up your Solr data is not strictly required (you can run knife index rebuild to get it back), but may make for a smoother recovery. Similarly, your rabbitmq config (generally in var/lib/rabbitmq/mnesia/ or similar) can be recreated (and Chef generally won’t touch it except in something like the bootstrap cookbook) but may be worth backing up as well.

And please do let us know about your experiences testing, both good and bad–it will help us decide when to go from beta to RC. For RCs, we’ve instituted a three day freeze policy where we wait for three days without a reported regression before shipping the final version.

Thanks in advance!


Dan DeLeo


#8

erm, how do I back up solr data? is in that in couch as well?

can I backup mnesia just by copying /var/lib/rabbitmq/mnesia? I have bad
memories of working w/ mnesia from ejabberd.

On Wed, Nov 2, 2011 at 7:13 AM, Daniel DeLeo dan@kallistec.com wrote:

On Tuesday, November 1, 2011 at 10:56 PM, Bryan Berry wrote:

what would be the best to protect my current system before testing it?

off the top of my head
back up couchdb
back up gems directory
backup /{var,usr}/lib/chef
stop chef services
put new chef code in place of old, depending on whether using gems or rpm
restart chef services

did I miss anything in this list?
Backing up your Solr data is not strictly required (you can run knife index rebuild to get it back), but may make for a smoother recovery.
Similarly, your rabbitmq config (generally in var/lib/rabbitmq/mnesia/ or
similar) can be recreated (and Chef generally won’t touch it except in
something like the bootstrap cookbook) but may be worth backing up as well.

And please do let us know about your experiences testing, both good and
bad–it will help us decide when to go from beta to RC. For RCs, we’ve
instituted a three day freeze policy where we wait for three days without a
reported regression before shipping the final version.

Thanks in advance!


Dan DeLeo


#9

+1 for live streaming.
-1 for certification. I dont kniw if certification really helps, at least
at this stage
chef server HA has been discusses various times in the mailing thread, as
well as in the IRC
would love to see discussion around the ticket triaging process.
Also how local chapters of chef community can be created/maintained and
more chef-hack days in different cities across the world (and how opscode
can enable/support that).

As of now, i prefer to off load the decision making of which chef server to
use (hosted or on premise ) on the clients. But i try to encourage adopting
opscode chef-server initially if the requirement is low (less than 5 nodes)
and if that adds value they can go ahead and change the engagement plan…
will love to hear how other are working,

On Wed, Nov 2, 2011 at 11:43 AM, Daniel DeLeo dan@kallistec.com wrote:

On Tuesday, November 1, 2011 at 10:56 PM, Bryan Berry wrote:

what would be the best to protect my current system before testing it?

off the top of my head
back up couchdb
back up gems directory
backup /{var,usr}/lib/chef
stop chef services
put new chef code in place of old, depending on whether using gems or rpm
restart chef services

did I miss anything in this list?
Backing up your Solr data is not strictly required (you can run knife index rebuild to get it back), but may make for a smoother recovery.
Similarly, your rabbitmq config (generally in var/lib/rabbitmq/mnesia/ or
similar) can be recreated (and Chef generally won’t touch it except in
something like the bootstrap cookbook) but may be worth backing up as well.

And please do let us know about your experiences testing, both good and
bad–it will help us decide when to go from beta to RC. For RCs, we’ve
instituted a three day freeze policy where we wait for three days without a
reported regression before shipping the final version.

Thanks in advance!


Dan DeLeo


#10

On Nov 2, 2011, at 2:08 AM, Ranjib Dey wrote:

+1 for live streaming.

I’m more on the fence. What would you stream? A single camera out in the lobby area? A single camera in each room? A camera every ten feet?

I might be interested in seeing videos of the presentations (with access to copies of the slides), but how much of the summit is going to be presentations versus discussions, bull sessions, and the “hallway track”? How much of that discussion might be totally irrelevant in a month or six?

-1 for certification. I dont kniw if certification really helps, at least at this stage

Agreed. Discussion regarding certification is appropriate for relatively mature products in a relatively mature space. Chef and DevOps are neither.

As of now, i prefer to off load the decision making of which chef server to use (hosted or on premise ) on the clients. But i try to encourage adopting opscode chef-server initially if the requirement is low (less than 5 nodes) and if that adds value they can go ahead and change the engagement plan… will love to hear how other are working,

IMO, a lot depends on who the clients are. For certain categories of clients, getting the lowest possible barrier to entry is key. For example, anyone who just wants to play around with it and see what they can get it to do for them, or who isn’t working in an industry where terms like SarbOx & PCI-DSS are used on a daily basis. Or anyone who already has most or all of their stuff “in the cloud”.

Installing and operating your own local chef-server is a non-trivial task with a fairly high barrier to entry, if all you’re going to do is use that to deploy and manage five nodes.


Brad Knowles bknowles@ihiji.com
SAGE Level IV, Chef Level 0.0.1


#11

On Wed, Nov 2, 2011 at 3:06 PM, Brad Knowles bknowles@ihiji.com wrote:

On Nov 2, 2011, at 2:08 AM, Ranjib Dey wrote:

+1 for live streaming.

I’m more on the fence. What would you stream? A single camera out in the
lobby area? A single camera in each room? A camera every ten feet?

the presentations and organized discussions

It would be nice to livestream the hallway track as well but I don’t think
that’s feasible

I might be interested in seeing videos of the presentations (with access
to copies of the slides), but how much of the summit is going to be
presentations versus discussions, bull sessions, and the “hallway track”?
How much of that discussion might be totally irrelevant in a month or six?

-1 for certification. I dont kniw if certification really helps, at
least at this stage

Agreed. Discussion regarding certification is appropriate for relatively
mature products in a relatively mature space. Chef and DevOps are neither.

As of now, i prefer to off load the decision making of which chef server
to use (hosted or on premise ) on the clients. But i try to encourage
adopting opscode chef-server initially if the requirement is low (less than
5 nodes) and if that adds value they can go ahead and change the engagement
plan… will love to hear how other are working,

IMO, a lot depends on who the clients are. For certain categories of
clients, getting the lowest possible barrier to entry is key. For example,
anyone who just wants to play around with it and see what they can get it
to do for them, or who isn’t working in an industry where terms like SarbOx
& PCI-DSS are used on a daily basis. Or anyone who already has most or all
of their stuff “in the cloud”.

Installing and operating your own local chef-server is a non-trivial task
with a fairly high barrier to entry, if all you’re going to do is use that
to deploy and manage five nodes.


Brad Knowles bknowles@ihiji.com
SAGE Level IV, Chef Level 0.0.1


#12

On Nov 2, 2011, at 2:08 AM, Ranjib Dey wrote:

Also how local chapters of chef community can be created/maintained and more chef-hack days in different cities across the world (and how opscode can enable/support that).

Thinking about this some more, and looking back to the local interactions we’ve had here on the topic of Chef in Austin, it seems to me that you can’t really isolate it from the broader context of DevOps, Agile Development, and Cloud Computing. It seems to me that you might be better off trying to foster the development of local groups in these areas, and try to keep the cross-pollination to a higher level.

Here in Austin, we have the Austin Cloud Users Group, AgileAustin, the Austin Hadoop Users Group, Cassandra Austin, LOPSA-Austin (LOPSA being the League of Professional System Administrators), the Austin Storage Networking User Group, CACTUS (the local USENIX chapter), various local Linux User Groups, and several SIGs organized underneath AgileAustin (including a DevOps SIG), and we have a lot of the same people who are active in multiple groups – including some of us who are members of the Chef community.

I think this cross-pollination is useful and important to maintain, and I’m not sure we could get enough people together in one place for a user group devoted to Chef exclusively.

DevOps does not operate by the old silo rules. By definition, it is cross-disciplinary. I don’t think it makes sense to try to cram DevOps back into a user group model that comes from the old-style everything-is-a-silo world.

I don’t think it makes sense to try to do that to Chef, either.


Brad Knowles bknowles@ihiji.com
SAGE Level IV, Chef Level 0.0.1


#13

“DevOps does not operate by the old silo rules. By definition, it is
cross-disciplinary. I don’t think it makes sense to try to cram DevOps
back into a user group model that comes from the old-style
everything-is-a-silo world.”

+1.

On Wed, Nov 2, 2011 at 9:21 AM, Brad Knowles bknowles@ihiji.com wrote:

On Nov 2, 2011, at 2:08 AM, Ranjib Dey wrote:

Also how local chapters of chef community can be created/maintained and
more chef-hack days in different cities across the world (and how opscode
can enable/support that).

Thinking about this some more, and looking back to the local interactions
we’ve had here on the topic of Chef in Austin, it seems to me that you
can’t really isolate it from the broader context of DevOps, Agile
Development, and Cloud Computing. It seems to me that you might be better
off trying to foster the development of local groups in these areas, and
try to keep the cross-pollination to a higher level.

Here in Austin, we have the Austin Cloud Users Group, AgileAustin, the
Austin Hadoop Users Group, Cassandra Austin, LOPSA-Austin (LOPSA being the
League of Professional System Administrators), the Austin Storage
Networking User Group, CACTUS (the local USENIX chapter), various local
Linux User Groups, and several SIGs organized underneath AgileAustin
(including a DevOps SIG), and we have a lot of the same people who are
active in multiple groups – including some of us who are members of the
Chef community.

I think this cross-pollination is useful and important to maintain, and
I’m not sure we could get enough people together in one place for a user
group devoted to Chef exclusively.

DevOps does not operate by the old silo rules. By definition, it is
cross-disciplinary. I don’t think it makes sense to try to cram DevOps
back into a user group model that comes from the old-style
everything-is-a-silo world.

I don’t think it makes sense to try to do that to Chef, either.


Brad Knowles bknowles@ihiji.com
SAGE Level IV, Chef Level 0.0.1


#14

On Wednesday, November 2, 2011 at 8:11 AM, Bryan Berry wrote:

On Wed, Nov 2, 2011 at 3:06 PM, Brad Knowles <bknowles@ihiji.com (mailto:bknowles@ihiji.com)> wrote:

I’m more on the fence. What would you stream? A single camera out in the lobby area? A single camera in each room? A camera every ten feet?

the presentations and organized discussions

It would be nice to livestream the hallway track as well but I don’t think that’s feasible

Just to clarify about the community summit, it will be an Open Spaces style “unconference” and not a more traditional conference with tracks, presentations and tutorials. Some of that will happen but it will be informal.


Opscode, Inc.
Joshua Timberman, Technical Program Manager
IRC, Skype, Twitter, Github: jtimberman


#15

On Tuesday, November 1, 2011 at 11:18 PM, Bryan Berry wrote:

erm, how do I back up solr data? is in that in couch as well?

It’s in the /var/chef directory by default (for gems at least).

can I backup mnesia just by copying /var/lib/rabbitmq/mnesia? I have bad memories of working w/ mnesia from ejabberd.
That should work.

Thanks again


#16

On Nov 2, 2011, at 7:59 AM, Joshua Timberman wrote:

It would be nice to livestream the hallway track as well but I don’t think that’s feasible

Just to clarify about the community summit, it will be an Open Spaces style “unconference” and not a more traditional conference with tracks, presentations and tutorials. Some of that will happen but it will be informal.

What if we also facilitate a virtual room that the far flung can join to participate in via google hangouts or the like?

sarah