pkgutil package provider (solaris)

[Rudy Gevaert]

Dear group,

Those who are running Solaris might be interested that there is some
movement to create a pkgutil provider so you can install packages from
opencsw.

Those who are interested, please see:
http://projects.puppetlabs.com/issues/4258

Thanks!

Rudy

[windowsrefund]

Are there any plans to retrofit this into 0.25.x?

[Nigel Kersten]

On Wed, Nov 3, 2010 at 8:18 AM, windowsrefund <window...@gmail.com> wrote:
> Are there any plans to retrofit this into 0.25.x?

I haven't looked at the actual provider yet, but you should be able to
distribute it yourself with pluginsync to 0.25.x clients once it's
been sorted.

We really need community testing with this provider. There are some
questions Rudy has in the ticket log that we need to get answered.

http://projects.puppetlabs.com/issues/4258

[Peter Bonivart]

On Nov 3, 4:39 pm, Nigel Kersten <ni...@puppetlabs.com> wrote:
> We really need community testing with this provider. There are some
> questions Rudy has in the ticket log that we need to get answered.
>
> http://projects.puppetlabs.com/issues/4258

Hi, I'm the author of pkgutil, I'm glad you're about to support it in
Puppet since I get quite many requests for that.

Unfortunately, I know nothing of Puppet or Python so I haven't been
able to help out my users there but if there's anything you want to
know about pkgutil or if you want something added/changed I will do my
best to help. The (much faster) --single option came from such a
request by Maciej Blizinski who contributed a new provider.

I think the issue with "Not installed" vs. "notinst" is probably from
converting the pkg-get provider, if I'm not mistaken pkg-get prints
"Not installed" so that should just be "notinst" instead. I'm thinking
about adding an option for machine parseable output to make these
things better.

I have changed the exit code to 0 (Rudy #10).

/peter

[Nigel Kersten]

Hi Peter, it's awesome to have you here.

We *love* machine parseable output in Puppet land :)

>
> I have changed the exit code to 0 (Rudy #10).
>
> /peter
>

> --
> You received this message because you are subscribed to the Google Groups "Puppet Users" group.
> To post to this group, send email to puppet...@googlegroups.com.
> To unsubscribe from this group, send email to puppet-users...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en.
>
>

--
Nigel Kersten - Puppet Labs -  http://www.puppetlabs.com

[Rudy Gevaert]

Hi Peter,

On Nov 12, 4:54 pm, Peter Bonivart <shuttle...@gmail.com> wrote:

> Hi, I'm the author of pkgutil, I'm glad you're about to support it in
> Puppet since I get quite many requests for that.

It's nice you chip in here! We appreciate it.

> I think the issue with "Not installed" vs. "notinst" is probably from
> converting the pkg-get provider, if I'm not mistaken pkg-get prints
> "Not installed" so that should just be "notinst" instead. I'm thinking
> about adding an option for machine parseable output to make these
> things better.

Pachine parseable output would be very nice, but in my opinion not
that 'urgent'. We can parse it like it is.

> I have changed the exit code to 0 (Rudy#10).

Nice. From which version would this be?

An other issue that we need to tackle is which version(s) of pkgutil
will we support with the puppet provider?

[Peter Bonivart]

On Nov 16, 11:17 am, Rudy Gevaert <rudy.geva...@gmail.com> wrote:
> Pachine parseable output would be very nice, but in my opinion not
> that 'urgent'.  We can parse it like it is.

I will add this as an option but not change the current output then.
Later, when you've had a chance to update the provider to use the
machine parseable output I can change the normal output more freely.

> > I have changed the exit code to 0 (Rudy#10).
>
> Nice.  From which version would this be?

I have it checked into SVN so it will be in v2.3 but I haven't set a
release date for it yet. Is it a showstopper for you? In that case I
could roll a v2.2.1 for you with this.

> An other issue that we need to tackle is which version(s) of pkgutil
> will we support with the puppet provider?

I really think you should use the --single option (as used in Maciej's
provider) for comparison due to performance reasons and that was in
v1.9, released roughly a year ago.

/peter

[Rudy Gevaert]

On Nov 16, 3:52 pm, Peter Bonivart <shuttle...@gmail.com> wrote:
> I have it checked into SVN so it will be in v2.3 but I haven't set a
> release date for it yet. Is it a showstopper for you? In that case I
> could roll a v2.2.1 for you with this.

Not for me personally. We will take this importunity to upgrade our
pkgutils on all the machines (not asap, but hopefully 'soon').

> > An other issue that we need to tackle is which version(s) of pkgutil
> > will we support with the puppet provider?
>
> I really think you should use the --single option (as used in Maciej's
> provider) for comparison due to performance reasons and that was in
> v1.9, released roug

Yes indeed. That is the minimum minimorum. However because of the
bug you just fixed, I think we could support the pkgutil provider
starting from v2.3.

[Dominic Cleal]

Hi Peter et al,

Sorry for arriving rather late to this thread!

On 16/11/10 10:17, Rudy Gevaert wrote:
> On Nov 12, 4:54 pm, Peter Bonivart <shuttle...@gmail.com> wrote:
>> I think the issue with "Not installed" vs. "notinst" is probably from
>> converting the pkg-get provider, if I'm not mistaken pkg-get prints
>> "Not installed" so that should just be "notinst" instead. I'm thinking
>> about adding an option for machine parseable output to make these
>> things better.
>
> Pachine parseable output would be very nice, but in my opinion not
> that 'urgent'. We can parse it like it is.

As Rudy said, we are able to parse the pkgutil output as-is, though it
would be great if there was a "quieter" mode for pkgutil as it can be
quite noisy. It's difficult to determine where the noise ends and the
package listings begin.

For example, with use_gpg checked, pkgutil outputs "Checking integrity"
messages and gpg itself outputs key information. If a catalog has to be
fetched (say catalog_update is 0, or expiry has been reached), then we
get information about which files are being fetched, plus wget output
(if -q isn't used).

I intend to start up a patch thread soon on puppet-dev with the combined
commits from James, Maciej, Rudy and me to begin the process of getting
the provider included.

Regards,

--
Dominic Cleal
Red Hat Consulting
m: +44 (0)7818 512168

[Frederic Conrotte]

Hello

Like many enterprises, we have an heterogeneous infrastructure with some CentOS and Solaris machines.

I was wondering if there are existing plans about creating for Solaris the equivalent of the "yumrepo" type ?

https://docs.puppetlabs.com/references/latest/type.html#yumrepo

Something like "pkgrepo" ?

Best regards

Frederic

[Wil Cooley]

On Oct 15, 2014 3:55 AM, "Frederic Conrotte" <frederic...@gmail.com> wrote:
>
> Hello
>
> Like many enterprises, we have an heterogeneous infrastructure with some CentOS and Solaris machines.
>

> I was wondering if lookthere are existing plans about creating for Solaris the equivalent of the "yumrepo" type ?

Unless this has changed in versions later than what I'm using, pkgutil only supports a single repository, so it's really just a matter of configuring pkgutil.conf. 'yumrepo' is useful because there can be any number of repos defined.

Wil