[Labs-l] [Ops] Debian Jessie now available on Labs

[Petr Bena]

Can I ask why you change from ubuntu to debian on production machines?

On Wed, Feb 18, 2015 at 5:28 AM, Andrew Bogott <abogott at wikimedia.org> wrote:
> That fact is known to be broken in recent versions of Facter;  if you
> encounter any code in puppet that relies on it, please change it to use our
> custom fact instead.
>
> For a lengthy elaboration on 'broken,' I refer you to Faidon's comments
> here:  https://gerrit.wikimedia.org/r/#/c/184303/
>
> -Andrew
>
>
> On 2/17/15 6:24 PM, Jeff Gage wrote:
>
> Thanks, this is great!
>
> I just ran into this difference with Facter: the version supplied with
> Jessie is newer and contains rewritten EC2 facts which don't seem to work in
> this image:
>
> $ lsb_release -c
> Codename: trusty
> $ dpkg -l facter | egrep '^ii' |  awk '{print $3}'
> 1.7.5-1ubuntu1
> $ dpkg -L facter | grep ec2
> /usr/lib/ruby/vendor_ruby/facter/ec2.rb
> /usr/lib/ruby/vendor_ruby/facter/util/ec2.rb
> $ sudo facter -p | egrep "^ec2_" | wc -l
> 17
>
> $ lsb_release -c
> Codename: jessie
> $ dpkg -l facter | egrep '^ii' |  awk '{print $3}'
> 2.2.0-1
> $ dpkg -L facter | grep ec2
> /usr/lib/ruby/vendor_ruby/facter/ec2.rb
> /usr/lib/ruby/vendor_ruby/facter/ec2/
> /usr/lib/ruby/vendor_ruby/facter/ec2/rest.rb
> /usr/lib/ruby/vendor_ruby/facter/util/ec2.rb
> $ sudo facter -p | egrep "^ec2_" | wc -l
> 0
>
> It's not clear to me what the cause of this regression is.
>
> In my case I was using the fact 'ec2_instance_id'. Luckily our custom fact
> /var/lib/puppet/lib/facter/ec2id.rb still works, so I switched my code to
> use 'ec2id' instead.
>
>
>
> On Tue, Feb 17, 2015 at 3:06 PM, Andrew Bogott <abogott at wikimedia.org>
> wrote:
>>
>> A new image type is now available for labs instances, 'debian-8.0-jessie'.
>> I've just now ironed out the last few bugs with this instance type, and it
>> should (finally!) be ready for long-term use.
>>
>> Late last year the WMF Operations team agreed that we would switch from
>> Ubuntu Trusty to Debian Jessie for new services.  It will be a very gradual
>> transition, but anyone who is developing new software (as opposed to working
>> on old, existing services) should go ahead and develop on a Jessie instance
>> so that you're ready for the future.
>>
>> There are a couple of small known issues with the original startup of
>> Jessie instances.  In some cases the very first puppet run will fail and
>> won't stabilize until the second run.  In very rare cases, an instance will
>> come up but fail to recognize the shared file systems in /data/public and
>> /home; if that happens a simple reboot should fix things.
>>
>> Please let me know if you encounter any unexpected issues with Debian
>> instances.
>>
>> -Andrew
>>
>>
>> _______________________________________________
>> Ops mailing list
>> Ops at lists.wikimedia.org
>> https://lists.wikimedia.org/mailman/listinfo/ops
>
>
>
>
> _______________________________________________
> Labs-l mailing list
> Labs-l at lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/labs-l
>