Mailing List Archive

Google changes to Quagga now available on code.google.com
Hi all;

I'd like to announce a newly set up git repo at code.google.com, where I
will be releasing changes by Google to Quagga - both features and fixes.

The first branch I've made available implements BGP multipath (by our own
Barry Friedman).

I will be announcing/adding further features in the coming weeks, and
we have plans in the works with our friends at the ISC to provide SQA
coverage for all these changes.

Here's a quick cheat sheet on how to obtain and build the BGP multipath
feature:

> $ git clone https://code.google.com/p/google-quagga/ bgp-multipath
> $ cd bgp-multipath
> $ git checkout bgp-multipath
> $ ./bootstrap.sh
> $ ./configure
> $ make

Any questions, please let me know!

Thanks,

--
Josh Bailey
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
On Thu, Jul 21, 2011 at 02:34:06PM -0700, Josh Bailey wrote:
> I'd like to announce a newly set up git repo at code.google.com, where I
> will be releasing changes by Google to Quagga - both features and fixes.
>
> The first branch I've made available implements BGP multipath (by our own
> Barry Friedman).

Hm, interesting, I only had time for a quick glaze but it seems your
implementation might not even break BGP completely... still, very touchy
subject there.

I'll definitely merge this into -dn42, we'll notice if it explodes
something.

> I will be announcing/adding further features in the coming weeks, and
> we have plans in the works with our friends at the ISC to provide SQA
> coverage for all these changes.

Oh yes, quagga direly needs QA.

> Any questions, please let me know!

Yeah, you wouldn't be hiring by chance? *g*


-David

_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
Thank you sir! More patches (and SQA results) on the way. If you've any
especial concerns where you think we could concentrate additional testing
effort please let me know.

On Wed, 27 Jul 2011, David Lamparter wrote:

> On Thu, Jul 21, 2011 at 02:34:06PM -0700, Josh Bailey wrote:
>> I'd like to announce a newly set up git repo at code.google.com, where I
>> will be releasing changes by Google to Quagga - both features and fixes.
>>
>> The first branch I've made available implements BGP multipath (by our own
>> Barry Friedman).
>
> Hm, interesting, I only had time for a quick glaze but it seems your
> implementation might not even break BGP completely... still, very touchy
> subject there.
>
> I'll definitely merge this into -dn42, we'll notice if it explodes
> something.
>
>> I will be announcing/adding further features in the coming weeks, and
>> we have plans in the works with our friends at the ISC to provide SQA
>> coverage for all these changes.
>
> Oh yes, quagga direly needs QA.
>
>> Any questions, please let me know!
>
> Yeah, you wouldn't be hiring by chance? *g*
>
>
> -David
>
>

--
Josh Bailey
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
Hi All,

New ISIS branch is pushed to google public repository.
http://code.google.com/p/google-quagga/source/browse/?name=is-is

This branch has support for new features and bug fixes.
Please let me know if you have any questions.

Here are the steps to enable ISIS:

> $ git clone https://code.google.com/p/**google-quagga/
> <https://code.google.com/p/google-quagga/>$ cd google-quagga
> $ git checkout is-is
> $ ./bootstrap.sh
> $ ./configure --enable-vtysh --enable-isisd --enable-multipath=0
> $ make
> $ make install


Special thanks to Josh for taking care of all the logistics.

Thank You,
--
Subbaiah Venkata
Technical Leader and Manager
Google Inc.
1600 Amphitheatre Parkway
Mountain View, CA 94043


On Thu, Jul 21, 2011 at 2:34 PM, Josh Bailey <joshb@google.com> wrote:

>
> Hi all;
>
> I'd like to announce a newly set up git repo at code.google.com, where I
> will be releasing changes by Google to Quagga - both features and fixes.
>
> The first branch I've made available implements BGP multipath (by our own
> Barry Friedman).
>
> I will be announcing/adding further features in the coming weeks, and we
> have plans in the works with our friends at the ISC to provide SQA coverage
> for all these changes.
>
> Here's a quick cheat sheet on how to obtain and build the BGP multipath
> feature:
>
> $ git clone https://code.google.com/p/**google-quagga/<https://code.google.com/p/google-quagga/>bgp-multipath
>> $ cd bgp-multipath
>> $ git checkout bgp-multipath
>> $ ./bootstrap.sh
>> $ ./configure
>> $ make
>>
>
> Any questions, please let me know!
>
> Thanks,
>
> --
> Josh Bailey
> ______________________________**_________________
> Quagga-dev mailing list
> Quagga-dev@lists.quagga.net
> http://lists.quagga.net/**mailman/listinfo/quagga-dev<http://lists.quagga.net/mailman/listinfo/quagga-dev>
>
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
01.02.2012, 13:52, "Subbaiah Venkata" <svenkata@google.com>:
> Hi All,
>
> New ISIS branch is pushed to google public repository.
> http://code.google.com/p/google-quagga/source/browse/?name=is-is
>
> This branch has support for new features and bug fixes.Please let me know if you have any questions.

Hello, Subbaiah.

Thank you for publishing this work. This is the second publication in the name of Google, and the final goal, as far as I take it, is to converge the branches in foreseeable future. It will require some work on the project's side as well.

I have briefly studied the contents of commits b1b73b9 and f17b498. First of all, proofreading is hard, because changes are not isolated from each other, the change to IS-IS, BGP, library functions, zserv and indentation are mixed together. Just in case the changes were originally made and verified in some ordered way and later joined in a single commit for the sole purpose of publication, discussing the original isolated changes one at a time would help making through the merge. Is it possible?

Then, there is a series of IS-IS commits, which noone was able to verify so far:

d83bf5a
30a7558
dfafeae
b775c38
9632e3e

The commit IDs belong to quagga-RE repo only. Do you have the time and will to review these? They seem to be very closely related.

--
Denis Ovsienko
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
On Fri, Feb 10, 2012 at 06:57:36PM +0400, Denis Ovsienko wrote:
> 01.02.2012, 13:52, "Subbaiah Venkata" <svenkata@google.com>:
> > Hi All,
> >
> > New ISIS branch is pushed to google public repository.
> > http://code.google.com/p/google-quagga/source/browse/?name=is-is
> >
> > This branch has support for new features and bug fixes.Please let me know if you have any questions.
>
> Then, there is a series of IS-IS commits, which noone was able to verify so far:

Considering that isisd is still "experimental", i would merge these with
"doesn't let hell break loose" as merge criterium.

-David
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
On 10/02/2012 15:11, David Lamparter wrote:
> Considering that isisd is still "experimental", i would merge these with
> "doesn't let hell break loose" as merge criterium.

+1

Nick
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
Hi Subbaiah Venkata,

On Wed, Feb 01, 2012 at 01:52:56AM -0800, Subbaiah Venkata wrote:
> New ISIS branch is pushed to google public repository.
> http://code.google.com/p/google-quagga/source/browse/?name=is-is
>
> This branch has support for new features and bug fixes.
> Please let me know if you have any questions.

I've just taken a look at the branch. Honestly, WTF? It comes as 2
patches with 518 / 130 hunks each, touching the entirety of isisd and
some files outside of isis?!

Please, please, split this into small reviewable patches. Patches that
touch things outside isisd/ need to be specially separated and explained
with proper comments.

Also, please do not mix whitespace/indent changes with code changes; if
you want to reindent code, that should be a separate patch that does
nothing else.


Cheers,


-David
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
Hi David;

Sorry about that - my fault. My intent here was to expedite the
availability of the code to the community and to the ISC for testing.

Clearly will need to come up with the right iterative approach to handle
large feature changes. Regarding the IS-IS changes specifically for
example. Would you be able to advise what areas we should prioritise for
repatching/commenting/for review? Idea of course is to come up with a
reusable process.

On Mon, 13 Feb 2012, David Lamparter wrote:

> Hi Subbaiah Venkata,
>
> On Wed, Feb 01, 2012 at 01:52:56AM -0800, Subbaiah Venkata wrote:
>> New ISIS branch is pushed to google public repository.
>> http://code.google.com/p/google-quagga/source/browse/?name=is-is
>>
>> This branch has support for new features and bug fixes.
>> Please let me know if you have any questions.
>
> I've just taken a look at the branch. Honestly, WTF? It comes as 2
> patches with 518 / 130 hunks each, touching the entirety of isisd and
> some files outside of isis?!
>
> Please, please, split this into small reviewable patches. Patches that
> touch things outside isisd/ need to be specially separated and explained
> with proper comments.
>
> Also, please do not mix whitespace/indent changes with code changes; if
> you want to reindent code, that should be a separate patch that does
> nothing else.
>
>
> Cheers,
>
>
> -David
>

--
Josh Bailey
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
Hi Denis,

On Fri, Feb 10, 2012 at 6:57 AM, Denis Ovsienko <infrastation@yandex.ru>wrote:

> 01.02.2012, 13:52, "Subbaiah Venkata" <svenkata@google.com>:
> > Hi All,
> >
> > New ISIS branch is pushed to google public repository.
> > http://code.google.com/p/google-quagga/source/browse/?name=is-is
> >
> > This branch has support for new features and bug fixes.Please let me
> know if you have any questions.
>
> Hello, Subbaiah.
>
> Thank you for publishing this work. This is the second publication in the
> name of Google, and the final goal, as far as I take it, is to converge the
> branches in foreseeable future. It will require some work on the project's
> side as well.
>

Agree, converging the branches is good for all of us in the community.


>
> I have briefly studied the contents of commits b1b73b9 and f17b498. First
> of all, proofreading is hard, because changes are not isolated from each
> other, the change to IS-IS, BGP, library functions, zserv and indentation
> are mixed together. Just in case the changes were originally made and
> verified in some ordered way and later joined in a single commit for the
> sole purpose of publication, discussing the original isolated changes one
> at a time would help making through the merge. Is it possible?
>

Quagga ISIS was not in a production ready state. I have touched almost all
parts of ISIS code and it take lot of time to cherry pick and get them
reviewed one at a time. Frankly there are 100s of changes. I think a
protocol code should go into slow moving small piece meal change reviews
when it gets into a maintenance mode, which ISIS is not in that state yet.
So I suggest we merge the code, test and commit to the head.


>
> Then, there is a series of IS-IS commits, which noone was able to verify
> so far:
>
> d83bf5a
> 30a7558
> dfafeae
> b775c38
> 9632e3e
>
> The commit IDs belong to quagga-RE repo only. Do you have the time and
> will to review these? They seem to be very closely related.
>

Yes, I have some time to review Quagga changes.


>
> --
> Denis Ovsienko
> _______________________________________________
> Quagga-dev mailing list
> Quagga-dev@lists.quagga.net
> http://lists.quagga.net/mailman/listinfo/quagga-dev
>



--
--
Subbaiah Venkata
Manager, Google Inc
1600 Amphetheatre Parkway
Mountain View, CA 94043
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
On Fri, Feb 10, 2012 at 7:11 AM, David Lamparter <equinox@diac24.net> wrote:

> On Fri, Feb 10, 2012 at 06:57:36PM +0400, Denis Ovsienko wrote:
> > 01.02.2012, 13:52, "Subbaiah Venkata" <svenkata@google.com>:
> > > Hi All,
> > >
> > > New ISIS branch is pushed to google public repository.
> > > http://code.google.com/p/google-quagga/source/browse/?name=is-is
> > >
> > > This branch has support for new features and bug fixes.Please let me
> know if you have any questions.
> >
> > Then, there is a series of IS-IS commits, which noone was able to verify
> so far:
>
> Considering that isisd is still "experimental", i would merge these with
> "doesn't let hell break loose" as merge criterium.
>

+1


>
> -David
> _______________________________________________
> Quagga-dev mailing list
> Quagga-dev@lists.quagga.net
> http://lists.quagga.net/mailman/listinfo/quagga-dev
>



--
--
Subbaiah Venkata
Manager, Google Inc
1600 Amphetheatre Parkway
Mountain View, CA 94043
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
On Mon, Feb 13, 2012 at 12:22:29PM -0800, Josh Bailey wrote:
>
> Hi David;
>
> Sorry about that - my fault. My intent here was to expedite the
> availability of the code to the community and to the ISC for testing.
>
> Clearly will need to come up with the right iterative approach to handle
> large feature changes. Regarding the IS-IS changes specifically for
> example. Would you be able to advise what areas we should prioritise for
> repatching/commenting/for review? Idea of course is to come up with a
> reusable process.

Aside from the obvious (creating small, reviewable, well-described
patches), I'd recommend sorting by "impact divided by semantic size"
factor. That would mean indentation/whitespace changes first (since they
have a semantic size of 0), then crash fixes (very large impact...),
then other bug fixes, etc.

I can't tell you how that boils down for isisd, my knowledge of this
part of Quagga is somewhat limited and I'm chronically short on time to
read code.

I would guess that the current interest in IS-IS is in part due to TRILL
and PBB SPB (802.1aq) both relying on IS-IS. If your employer's interest
is in TRILL, I would heavily recommend taking a look at
http://src.illumos.org/source/xref/illumos-userland/components/quagga/patches/
(in particular 70-isisd-trill.patch)

It would also make sense to submit patches as you develop them, not in
one giant batch. It is far easier to merge a patch or two per day over a
month than a pile of 50 in one go. But, alas, quagga maintainership
still works in a very ad-hoc pseudorandom "i need this and that" basis,
so getting /anything/ merged depends on available time and fancy of very
few people.


Cheers,


-David
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
> Considering that isisd is still "experimental", i would merge these with
> "doesn't let hell break loose" as merge criterium.

Done.

--
Denis Ovsienko
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
On Thu, 21 Jul 2011, Josh Bailey wrote:

> Any questions, please let me know!

Have you seen:

https://datatracker.ietf.org/doc/draft-ietf-idr-deprecate-as-sets/

and what do you think is the way forward to deal with this (get IDR to
reconsider deprecating AS_SETs, try use Add-Path instead in some way)?

regards,
--
Paul Jakma paul@jakma.org twitter: @pjakma PGP: 64A2FF6A
Fortune:
Ginsburg's Law:
At the precise moment you take off your shoe in a shoe store, your
big toe will pop out of your sock to see what's going on.
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
On Mon, Mar 5, 2012 at 06:11, Paul Jakma <paul@jakma.org> wrote:
> On Thu, 21 Jul 2011, Josh Bailey wrote:
>
>> Any questions, please let me know!
>
>
> Have you seen:
>
>  https://datatracker.ietf.org/doc/draft-ietf-idr-deprecate-as-sets/
>
> and what do you think is the way forward to deal with this (get IDR to
> reconsider deprecating AS_SETs, try use Add-Path instead in some way)?

Paul,

I reread this thread and I'm not sure what is driving your question.
As I know Warren Kumari, one of the authors of the draft, I CCed him
directly. Can you elaborate a bit on what you see needing to be dealt
with?

Thanks,
Scott

>
> regards,
> --
> Paul Jakma  paul@jakma.org  twitter: @pjakma  PGP: 64A2FF6A
> Fortune:
> Ginsburg's Law:
>        At the precise moment you take off your shoe in a shoe store, your
>        big toe will pop out of your sock to see what's going on.
>
> _______________________________________________
> Quagga-dev mailing list
> Quagga-dev@lists.quagga.net
> http://lists.quagga.net/mailman/listinfo/quagga-dev



--
panem et circenses - a winning strategy for over 2000 years!

_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev
Re: Google changes to Quagga now available on code.google.com [ In reply to ]
Hi Scott,

On Tue, 6 Mar 2012, Scott Whyte wrote:

> I reread this thread and I'm not sure what is driving your question. As
> I know Warren Kumari, one of the authors of the draft, I CCed him
> directly. Can you elaborate a bit on what you see needing to be dealt
> with?

I had a look through the code. If I understand correctly it creates an
aggregate to deal with merging routes where the attributes differ (a good
thing, in my book). So my question is how this fits in IDR having
deprecated the AS_SET segment type in AS_PATH attribute (which what is
used to merge together the ASNs)?

This doesn't have any material affect on the changeset, for now. I'm just
wondering have you thought about the future, and what happens should
AS_SET go from just deprecated to not-supported-at-all by any widely-used
future implementation?

regards,
--
Paul Jakma paul@jakma.org @pjakma Key ID: 64A2FF6A
Fortune:
The universe does not have laws -- it has habits, and habits can be broken.
_______________________________________________
Quagga-dev mailing list
Quagga-dev@lists.quagga.net
http://lists.quagga.net/mailman/listinfo/quagga-dev