Questions for DPL candidates: Support channels

[Ritesh Raj Sarraf]

Hello DPL Candidates,

Thank you for standing for the DPL position.

Debian as a project is different than others. Most other similar projects, have
a commercial backing and interest. This puts the onus on them (other Linux
distributions) to ensure their support infrastructure is simple, intuitive and
supportable.

Debian, on the other hand, is limited by community support. Community Support
being, "Best Effort from the Community". Also, many of the Debian Community
Support channels are fragmented (IRC, ML, Ask, Forum, Web)

Do you think the current support model/system in place, is good enough ?
Do we need (or have room for) a different approach ?


Thanks,
Ritesh

[Chris Lamb]

Hello Ritesh,

Post by Ritesh Raj Sarraf
Debian as a project is different than others. Most other similar projects, have
a commercial backing and interest. This puts the onus on them (other Linux
distributions) to ensure their support infrastructure is simple, intuitive and
supportable.

You raise some interesting points especially regarding fragmentation of
support. To be fair to Debian, even projects with "commercial backing" have
this issue. :)

I think we should judge ourselves on our results and not whether X or Y
exists. In other words, it doesn't matter whether the support is operated by
paid operatives or by the community, the key question is whether our users
are acutally getting the support they need.

However, it is very difficult to get concrete answers here. Anecdotes are
not data, but if we hear enough times that "Yeah, I tried using Debian but
my wifi/video/keyboard/smartcard didn't work, but it worked under Z…" then
we might start to question whether we are serving our users best.

Somewhat related; as I outlined in my platform, I intend to perform some
usability testing to identify our biggest bottlenecks in known problem
areas.

In general, I think we are doing *pretty* well and specific targeting of
managable support concerns (eg. security support for oldstable ie. Debian
LTS) have been shown to be effective. However, we will always be limited by
the number of hours we can dedicate to Debian on a volunteer basis.

Correct or not, providing support also has a reputation for not being very
rewarding, so incentivising developers to spend will regrettably remain an
uphill battle.


Regards,

[Ritesh Raj Sarraf]

hello Chris,

Thank you for your response. Please see below some follow-ups.

Post by Chris Lamb
Hello Ritesh,

Post by Ritesh Raj Sarraf
Debian as a project is different than others. Most other similar projects, have
a commercial backing and interest. This puts the onus on them (other Linux
distributions) to ensure their support infrastructure is simple, intuitive and
supportable.

You raise some interesting points especially regarding fragmentation of
support. To be fair to Debian, even projects with "commercial backing" have
this issue. :)
I think we should judge ourselves on our results and not whether X or Y
exists. In other words, it doesn't matter whether the support is operated by
paid operatives or by the community, the key question is whether our users
are acutally getting the support they need.
However, it is very difficult to get concrete answers here. Anecdotes are
not data, but if we hear enough times that "Yeah, I tried using Debian but
my wifi/video/keyboard/smartcard didn't work, but it worked under Z
" then
we might start to question whether we are serving our users best.

I admin "Support" is a much wider topic. So I'll take examples to phrase my
questions.

Most major hardware and software vendors have a handful of Linux distributions
marked as supported, even though there's a high possibility that other
distributions would work equally well. I understand marking an item "Supported"
has other challenges, which is what leaves out most of the non-commercial
distributions. 

I myself, with one of my previous employer, had challenges mentioning support.
Back then, we rather chose to cook up a "Community Support Model" for Debian and
similar non-commercial* GNU/Linux distributions.

ISVs have improved over time. Today, 2 of the tools that I use (Crossover and
Skype), do offer a Debian .deb package.

On the IHV front, I am not sure if things are the same:
* We have a very small list of IHVs [1] mentioning support for Debian.
* We have a Debian Enterprise Mailing List [2]. But I don't see much traffic
there


Do you think a HW Certification Process should be available for Debian ?
I see all other Enterprise Distributions (including Debian Derivatives) have a
certification process in place.

Should we have a Certification Test Suite (Both Hardware and Software) that our
vendors could run and provide us with the results ? 
Such results could be validated by our Enterprise Team, and accordingly a HCL
could be set up for Debian.



[1] https://wiki.debian.org/Hardware/ShippingWithDebian
[2] https://lists.debian.org/debian-enterprise/

* Non Commercial as in not backed by a profit making organization

[Chris Lamb]

Post by Ritesh Raj Sarraf
Do you think a HW Certification Process should be available for Debian ?

[…]

Post by Ritesh Raj Sarraf
Should we have a Certification Test Suite

As Project Leader I would not stand in the way of anyone wishing to
work on this and can certainly see value to Debian if they existed.
However, at this point I remain unconvinced that this is would be
the most valuable use of a Leader's attention throughout a term.

This is not because getting commercial vendors interested in Debian
is of no value (!), I just believe that developer-driven or grassroots
"bottom-up" efforts are just more effective at increasing our
userbase.

For example, getting VPS providers to switch their defaults to Debian
or for hackable IoT devices to ship with Debian would cause trickle-up
effects into the enterprise space over time, without the need for a
perceived-bureaucratic processes of hardware signoff or hardware
compatibility reporting.


Best wishes,

[Paul Wise]

Post by Ritesh Raj Sarraf
Do you think a HW Certification Process should be available for Debian ?

I note that some hardware vendors are asking for one, one example here:

https://lists.debian.org/msgid-search/***@MAILBX03.quanta.corp

[Thomas Goirand]

Post by Ritesh Raj Sarraf
Do you think a HW Certification Process should be available for Debian ?

Even more, from my experience, the availability of an HCL (ie: Hardware
Compatibility List) is mandatory for some vendors to choose Debian. At
$work, I've been told that Debian wouldn't an option for it.

I had thoughts about this, and the only way I see it happening, is by
having some kind of software comparable to Ubuntu's "Checkbox", with
some report on our website. Then through this process, offer vendors the
possibility to put a "Debian compatible" logo on their website. To avoid
trademark issues, this logo *must* be different from the standard Debian
logo.

Note that I am convince the FSF initiative (I can't remember the ugly
website name, sorry...) is *not* a correct response to this need. We
need our own thing.

If one was to start such initiative, I'd be happy to help.

Cheers,

Thomas Goirand (zigo)

[Paul Wise]

Post by Thomas Goirand
Even more, from my experience, the availability of an HCL (ie: Hardware
Compatibility List) is mandatory for some vendors to choose Debian. At
$work, I've been told that Debian wouldn't an option for it.

Would this count as a HCL for your $work?

https://wiki.debian.org/InstallingDebianOn

Post by Thomas Goirand
I had thoughts about this, and the only way I see it happening, is by
having some kind of software comparable to Ubuntu's "Checkbox", with
some report on our website. Then through this process, offer vendors the
possibility to put a "Debian compatible" logo on their website. To avoid
trademark issues, this logo *must* be different from the standard Debian
logo.

For reference:

https://help.ubuntu.com/community/Checkbox
https://wiki.ubuntu.com/Testing/Automation/Checkbox

Some related projects:

https://linux-hardware.org/
https://github.com/linuxhw/hw-probe (warning, quite invasive)

https://en.wikipedia.org/wiki/Smolt_(Linux)
https://en.opensuse.org/Smolt
http://bugs.debian.org/435058
https://fedoraproject.org/wiki/Smolt_retirement

Post by Thomas Goirand
Note that I am convince the FSF initiative (I can't remember the ugly
website name, sorry...) is *not* a correct response to this need. We
need our own thing.

I guess you mean one of these two:

https://www.fsf.org/resources/hw/endorsement/respects-your-freedom
https://h-node.org/

[Ritesh Raj Sarraf]

Post by Paul Wise

Post by Thomas Goirand
Even more, from my experience, the availability of an HCL (ie: Hardware
Compatibility List) is mandatory for some vendors to choose Debian. At
$work, I've been told that Debian wouldn't an option for it.

Would this count as a HCL for your $work?
https://wiki.debian.org/InstallingDebianOn

I don't think any data, fed in good faith, could be termed as an HCL input.
We should eye for certification tools similar to what Red Hat, SUSE, Canonical
and Oracle may be using.

In quick search, I came across:
http://www.oracle.com/webfolder/technetwork/hcl/hcts/index.html
which looks similar to what I've wished for Debian.

Unexplored results showed the following at the very first result.
https://linux-test-project.github.io/
This one emphasized on Linux and if it serves the purpose, it could be a great
start for Debian, because Linux is our prime release target. But eventually,
we'd want to have something which could also cover parts of BSD and Hurd too.


Maybe if nothing suits, we could leverage GSoC and come up with something
tailored to the projects scope.

But, the biggest question is if we all (as project members) see, in unison, the
need for such a certification list.


I think this question and the subject fits in in the broader topic of
"universality".

[Ritesh Raj Sarraf]

Post by Ritesh Raj Sarraf

Post by Paul Wise
Would this count as a HCL for your $work?
 
https://wiki.debian.org/InstallingDebianOn

I don't think any data, fed in good faith, could be termed as an HCL input.
We should eye for certification tools similar to what Red Hat, SUSE, Canonical
and Oracle may be using.

A process similar to this is what would be useful to our project, in my opinion.
Picked from Red Hat's HCL Documentation [1]

* The partner establishes a certification relationship with Red Hat.
* The partner creates a request for the certification of a specific system or
hardware component on Red Hat's internal certification website (https://hardware
.redhat.com).
Red Hat's certification review team creates an official test plan inside the
certification request to track the testing of all relevant components.
* The partner runs the tests specified in the official test plan and submits
results packages to Red Hat for analysis.
* The review team analyzes the test results and marks them as completed when
they receive passing results. Any failures require retesting.
* The partner provides Red Hat with a representative hardware sample that covers
the items that are being certified.
* The system or component is marked as certified when all tests have passing
results, and the entry is made visible to the public on the external Red Hat
Hardware Certification website at https://access.redhat.com/certifications if
the partner requested a published certification.

[1] https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux_Hardware_Certification/1/html/Test_Suite_User_Guide/sect-User_Guide-Certification_Process_Overview-Summaries.html

[Thomas Goirand]

Post by Ritesh Raj Sarraf
But, the biggest question is if we all (as project members) see, in unison, the
need for such a certification list.

We don't need that *all* DDs see this need. If a few see it, and act to
make it happen, and at the same time, it doesn't bother the other DDs,
then we're fine.

Cheers,

Thomas Goirand (zigo)

[Thomas Goirand]

Post by Paul Wise

Post by Thomas Goirand
Even more, from my experience, the availability of an HCL (ie: Hardware
Compatibility List) is mandatory for some vendors to choose Debian. At
$work, I've been told that Debian wouldn't an option for it.

Would this count as a HCL for your $work?
https://wiki.debian.org/InstallingDebianOn

No, because it's just a wiki entry, and there's no way to check how it
has been written. A well established process needs to be written
(comparable to the Red Hat one that Ritesh wrote later in this thread).

Post by Paul Wise

Post by Thomas Goirand
Note that I am convince the FSF initiative (I can't remember the ugly
website name, sorry...) is *not* a correct response to this need. We
need our own thing.

https://www.fsf.org/resources/hw/endorsement/respects-your-freedom
https://h-node.org/

Yes, I was referring to h-node. I've been told multiple times that
h-node was the HCL for Debian. First, the site is really ugly, and
doesn't look serious. Then when you click on hardware, there's a list of
"stuff" (scanners, webcams, etc.) but ... no servers! This really looks
like laptop oriented to me.

Cheers,

Thomas Goirand (zigo)

[Mehdi Dogguy]

Hi Ritesh,

Post by Ritesh Raj Sarraf
Hello DPL Candidates,
Thank you for standing for the DPL position.
Debian as a project is different than others. Most other similar projects, have
a commercial backing and interest. This puts the onus on them (other Linux
distributions) to ensure their support infrastructure is simple, intuitive and
supportable.
Debian, on the other hand, is limited by community support. Community Support
being, "Best Effort from the Community". Also, many of the Debian Community
Support channels are fragmented (IRC, ML, Ask, Forum, Web)
Do you think the current support model/system in place, is good enough ?

Compared to other FLOSS projects, it is quite good. I have recently attended
a meetup (not about Debian) and we ended up discussing about community support
and comparing between a few projects. It turns out that people find our support
channels to be welcoming and helpful. So kudos to those who try to help users
everyday!

Post by Ritesh Raj Sarraf
Do we need (or have room for) a different approach ?

(The discussion turned more on certification programs. So I'd focus on that
aspect in my reply).

From a user and corporate perspectives, it is always nice to check if some
hardware is well supported in Debian (accross releases). As noted in this
thread, the lack of a certification program is indeed a blocker for some
vendors. It would help everyone to have a compatibility list. I'd support
such initiative, of course, if there are volunteers for this. I have to
admit I am not particularly familiar with how similar program usually
work but I am happy to trust DDs motivated to work on this.

If we have a clear plan, and like for every certification lab, it would be
very easy to convince manufacturers to give us some hardware in order to check
its compatibility and report useful bugs when something is missing or
ill-working.

Finally, I do not imagine any DPL (or even DD) to block such initiatives. So
a better way to start this could be to:
1) Write down a description of the certification program you want for Debian
2) Potentially, work a list of actions to implement it
3) (Collectively) think about how the project can make an official statement
about the certification program
4) Discuss #1, #2 and #3 on a public mailing-list to gather input from other
project members.

Regards,