CentOS - Dec 2015 - wifi on servers and fedora [was Re: 7.2 kernel panic on boot]

Warren Young wrote:
> On Dec 9, 2015, at 11:55 AM, m.roth at 5-cent.us wrote:
>> Matthew Miller wrote:
>>> On Wed, Dec 09, 2015 at 01:05:15PM -0500, m.roth at 5-cent.us
wrote:
>>>>>>
>>>> So, you're saying that end users need to go poke their
noses into the
>>>> development process
>>>
>>> If you want to go out of your way to read it that way, it's
hard to
>>> stop you. However, it's not what I'm saying. The
development process is
>>> conducted in the open for a reason.
>>
>> I don't see that as going "out of my way". Let's put
it this way: how
>> many times have folks on the development side poked their nose in here
>> - the general redhat list is pretty dead - and asked anything?
>
> So?you want veto power over Fedora?  You want every proposed change to
> cross your desk for a yea/nay?
Beg pardon? Why are you caricaturing what I said? I don't believe any of
us who are complaining are talking about every small change; rather, the
major ones.

As a lesser example, I just *adore* the new ethernet names - NOT. Breaks
scripts, makes it all more difficult, not to mention *so* much easier to
guess, when you've debugging a box and your organization has hardware from
many OEMs. What was wrong with eth0, or even em1? Why go to Sun naming
conventions? Maybe it helps EEs, but not sysadmins.

Please, though, that naming is *not* the point of the
thread.
>
> What if the Fedora project gatewayed the low-traffic development mailing
> list to this one, so that you don?t even have *that* barrier to
> participation?  Now ask yourself: what user-visible changes do you expect
> in the world afterward?
Why not what was suggested, a summary every month or three? How about
sending announcements?
<snip>
> People give Poettering a lot of static, but the fact is, he Gets. Stuff.
> Done.  If you want different stuff done, you?re going to have to make that
> happen somehow.  Shouted complaints from a soapbox don?t compile.
Which a vast number of us strongly opposed, but were not listened to. That
stuff is fine for a desktop, but who *cares* How Fast a *server* Shuts
Down? And coming up - hell, damn HP server take for-bloody-*ever* with
their firmware, init V is faster than their firmware.
>
> And don?t play the ?underfunded government agency? card.  LANL, LLBL,
> ORNL, NASA, USGS?all have given back lots of code to the open source
> world.  As well they should, because they derive an awful lot of benefit
> from that world.
May be, but my federal agency is at *least* 5% under what we were getting
in 2003, and my manager, who's working with another Institute about 2/3rds
of his time, and I, and another admin have to manage over 170 servers,
workstations, and clusters, some with special software, and ranging in age
from just bought to 2007 (I think there may be a workstation or 3 older),
and some of which we haven't managed to get the owners to allow us to get
off CentOS 5....
>
> I?m not against your basic position, Mark.  I, too, have shaken my head in
> dismay at several of the desktop-focused behaviors in recent versions of
> CentOS.[*]  I think where we actually differ is that I realize that I have
> no right to complain all that loudly about them, because I have the means
> to change them, but do not.
And I ask permission from my fed manager to put in a ticket with upstream
(which reminds me, I need to ask about putting one in for those docs with
links to google ads).
>
> Partly that?s because of differing priorities, partly it?s out of rational
> self-interest (i.e. I know how many OS forks fizzle) and yes, it?s partly
> just laziness.  But there?s that difference: I know why I?m not out there
> trying to change it.
>
> What are your reasons?
>
Lack of time, as I've indicated.
>
> [*] My favorite fumble is the one where a 2-NIC box with one DHCP
> interface and one static will swap the configurations silently when you
> boot with only the DHCP cable plugged in.  Because *obviously* you want
> the static IP to be available all the time, right?  This is great for wifi
> + Ethernet laptops, where you want the static IP to move when you plug the
> wired LAN cable in, but it doesn?t work out so great for servers where the
> DHCP NIC is normally disconnected, and exists only so the boots on the
> ground can move the cable in an emergency to reestablish the Internet link
> after they roached the LAN config somehow.  This behavior means the broken
> static IP moves to the secondary NIC, where it remains broken.  Solution:
> Plug both network cables in so NetworkManager doesn?t get Clever.?
Oh, I remember when you couldn't be sure, pre-NM, what would be eth0,
until you put the MAC address in....

       mark

On Thu, Dec 10, 2015 at 04:56:34PM -0500, m.roth at 5-cent.us
wrote:
> Why not what was suggested, a summary every month or three? How about
> sending announcements?
Do people _want_ accepted Fedora change announcements posted to this
list? That's pretty easy to arrange if it really helps. I don't see a
big benefit over just following the annoucement list where they're
posted (filtering out other topics if you want), but if people would
really find that helpful, we could do it.

-- 
Matthew Miller
<mattdm at fedoraproject.org>
Fedora Project Leader

On 12/10/2015 1:56 PM, m.roth at 5-cent.us wrote:
> As a lesser example, I just*adore*  the new ethernet names - NOT. Breaks
> scripts, makes it all more difficult, not to mention*so*  much easier to
> guess, when you've debugging a box and your organization has hardware
from
> many OEMs. What was wrong with eth0, or even em1?
when you have multiple adapters, perhaps different types (maybe 2 10gigE 
and 2 1gigE?) which one is eth0 supposed to be?   BSD has always used 
driver type in the network device names, and having dealt with device 
confusions before, I understand why.



-- 
john r pierce, recycling bits in santa cruz

On 12/10/2015 2:32 PM, Matthew Miller wrote:
> On Thu, Dec 10, 2015 at 04:56:34PM -0500,m.roth at 5-cent.us  wrote:
>> >Why not what was suggested, a summary every month or three? How
about
>> >sending announcements?
> Do people_want_  accepted Fedora change announcements posted to this
> list? That's pretty easy to arrange if it really helps. I don't see
a
> big benefit over just following the annoucement list where they're
> posted (filtering out other topics if you want), but if people would
> really find that helpful, we could do it.
I personally say 'NO' to this.  I have zilch interest in Fedora.


-- 
john r pierce, recycling bits in santa cruz

You think this is irritating, what about when you're trying to replicate the
network configuration to failover hardware...  There is a way around this, I
haven't tried it on CentOS but on Ubuntu there are kernel command line
parameters:

net.ifnames=1
biosdevname=0

which will override this behavior. Again, on Ubuntu these are added in
/etc/default/grub as parameters to GRUB_CMDLINE_LINUX_DEFAULT.  Finally,
there's /udev/rules.d/70-persistent-net.rules which allows you to associate
a MAC address with an eth? label.  However, without the command line parameters
it is ignored (contrary to other statements on the web ).  Given this is CentOS
you mileage will almost certainly vary but hopefully this gives you enough to go
on to get to the final solution.  There is a freedesktop.org web page about why
they did this - it has to do with mobile devices and plug-and-play networking. 
Take that page's statement about setting net.ifnames=0 cautiously, I found
it was the exact opposite.  biosdevname is a program written by someone at Dell
which is supposed to report on hardware configurations and make some sense out
of the cesspool.  It appears the source of the whole thing is hardware vendors
doing whatever they want and in some cases not playing by the rules.

----- Original Message -----
From: "John R Pierce" <pierce at hogranch.com>
To: centos at centos.org
Sent: Thursday, December 10, 2015 4:33:24 PM
Subject: Re: [CentOS] wifi on servers and fedora [was Re: 7.2 kernel panic on
boot]

On 12/10/2015 1:56 PM, m.roth at 5-cent.us wrote:
> As a lesser example, I just*adore*  the new ethernet names - NOT. Breaks
> scripts, makes it all more difficult, not to mention*so*  much easier to
> guess, when you've debugging a box and your organization has hardware
from
> many OEMs. What was wrong with eth0, or even em1?
when you have multiple adapters, perhaps different types (maybe 2 10gigE 
and 2 1gigE?) which one is eth0 supposed to be?   BSD has always used 
driver type in the network device names, and having dealt with device 
confusions before, I understand why.



-- 
john r pierce, recycling bits in santa cruz

_______________________________________________
CentOS mailing list
CentOS at centos.org
https://lists.centos.org/mailman/listinfo/centos

Confidentiality Notice | This email and any included attachments may be
privileged, confidential and/or otherwise protected from disclosure.  Access to
this email by anyone other than the intended recipient is unauthorized.  If you
believe you have received this email in error, please contact the sender
immediately and delete all copies.  If you are not the intended recipient, you
are notified that disclosing, copying, distributing or taking any action in
reliance on the contents of this information is strictly prohibited.

On Dec 10, 2015, at 2:56 PM, m.roth at 5-cent.us wrote:
> 
> Warren Young wrote:
>> So?you want veto power over Fedora?
> 
> Beg pardon? Why are you caricaturing what I said?
I didn?t think it was a caricature at all.  You clearly don?t want people to
?listen? to you, you want veto power.

If all you wanted was to be heard, you?d have stopped banging on this drum long
ago.  We got it.  We heard you.  You don?t like it.

How else would you characterize a desire for wishes to be changes, other than
veto power?
> As a lesser example, I just *adore* the new ethernet names - NOT. Breaks
> scripts
Hard-coded values are never a good idea.  That?s been a principle of good
software design and systems administration since the 1960s, at least.

The outputs of ip link and ifconfig -a are parseable for a reason.  Or, you can
iterate over the contents of /sys/class/net.

Mind, I didn?t come away from that change unscathed.  I had to go back and make
some changes to my code.  I think it amounted to about an hour of work, done
years ago, and amortized to all-but-zero since then.

The bigger problem is the day-to-day mystery of it all.  ?Gee, Brain, what
interface shall we bounce tonight??  ?The same interface we bounce every night:
enp3s0!?  ?But Braaain, it?s been called enp4s0 ever since the mobo manufacturer
switched to the rev 2 boards!  Narf!?  15 minutes of comic violence later,
followed by utter failure; then, ?So, Brain, what shall we do tomorrow night?? 
?The same thing we do every night, Pinky: try to bounce the first Ethernet NIC!?
>> What if the Fedora project gatewayed the low-traffic development
mailing
>> list to this one, so that you don?t even have *that* barrier to
>> participation?  Now ask yourself: what user-visible changes do you
expect
>> in the world afterward?
> 
> Why not what was suggested, a summary every month or three? How about
> sending announcements?
Fine, I repeat my question: what user-visible change do you expect to find in
the world after they do that, given that those receiving only those
announcements (i.e. those not also watching the Fedora dev lists) will
contribute precisely *squat* other than complaints?

Once again, soapbox soliloquies don?t compile.
> <snip>
>> People give Poettering a lot of static, but the fact is, he Gets.
Stuff.
>> Done.  If you want different stuff done, you?re going to have to make
that
>> happen somehow.
> 
> Which a vast number of us strongly opposed
Opposed what, exactly?  Everything Poettering has ever done, or did you have
something specific in mind?
> but were not listened to.
I took a wild guess that your complaints are about systemd, rather than avahi,
pulseaudio, or any of the other several dozen projects Lennart Poettering has
worked on.

I got 210 results from Googling CentOS?s mailing list archive server for your
email address and ?systemd?.  The first one appeared in 2014, *four years* after
systemd was created, and over three years after it was released as the default
init system for Fedora.

And that was the *only* post from you on that topic in 2014.  The other 209
posts were all in 2015, when it was way, way too late to change the decision.

So, in what world do your 2015 wishes for systemd to go away become a change in
that world?
> who *cares* How Fast a *server* Shuts
> Down? And coming up - hell, damn HP server take for-bloody-*ever* with
> their firmware, init V is faster than their firmware.
We?ve covered this already: the cloud cares.

It?s right there on the front page of https://www.digitalocean.com/  They can
bring a VM up for you in 55 seconds.  How do you suppose they achieved that?

It isn?t just one company?s marketing slogan.  Rackspace, Amazon, etc., all
start from a few key premises, one of which is that you can spin a server up and
down fast enough that you can rent dynamic instant-to-instant slices of the host
hardware, as opposed to the old VPS or shared hosting models, where the finest
rental time granularity was a month.

This is a multi-billion dollar business.[1]  You can?t handwave it away as
unimportant.  Red Hat would have to be fools not to be running hard to grab a
slice of that pie.


[1]: http://www.bbc.co.uk/news/business-32442268
>> And don?t play the ?underfunded government agency? card.  LANL, LLBL,
>> ORNL, NASA, USGS?all have given back lots of code to the open source
>> world.  As well they should, because they derive an awful lot of
benefit
>> from that world.
> 
> May be, but my federal agency is at *least* 5% under what we were getting
> in 2003
Sigh?so you go and play the card anyway.

What, you think NASA?s doing great?  Their operating budget was about 1/20 that
spent on troops? air conditioning in Iraq and Afghanistan in 2011.[2]

Maybe you think the national labs are flush with cash, here in the post cold war
era?

Open source works on the stone soup principle: everyone goes hungry when they
hang onto their gnarled carrots and wrinkled potatoes, but everyone has stew
when they throw their scraps into the pot.[3]


[2]: https://goo.gl/WejcWK
[3]: https://goo.gl/t2pSMw
>> Partly that?s because of differing priorities, partly it?s out of
rational
>> self-interest (i.e. I know how many OS forks fizzle) and yes, it?s
partly
>> just laziness.  But there?s that difference: I know why I?m not out
there
>> trying to change it.
>> 
>> What are your reasons?
>> 
> Lack of time, as I've indicated.
By demanding changes to the software without contributing patches to effect
those changes, you are in effect imposing a burden on other peoples? time.

We have a model for that: I pay you $X for Y hours of time, so that I do not
have to spend the Y hours myself, or acquire the specialized knowledge it takes
to apply those hours productively.  You can do that in the form of an employment
contract, or a support contract, or a donation.

If you will neither contribute your own hours nor pay for someone else?s hours,
all you?ve got left is soapboxing.
>> Plug both network cables in so NetworkManager doesn?t get Clever.?
> 
> Oh, I remember when you couldn't be sure, pre-NM, what would be eth0,
> until you put the MAC address in?.
I thought the MAC address binding solution was a perfectly reasonable way to
solve the problem, and I want it back.

The problem now is that NM in EL7 treats the MAC address as a mere hint, rather
than a command.

On 12/10/2015 05:33 PM, John R Pierce wrote:
> On 12/10/2015 1:56 PM, m.roth at 5-cent.us wrote:
>> As a lesser example, I just*adore*  the new ethernet names - NOT.
Breaks
>> scripts, makes it all more difficult, not to mention*so*  much easier
to
>> guess, when you've debugging a box and your organization has
hardware from
>> many OEMs. What was wrong with eth0, or even em1?
> when you have multiple adapters, perhaps different types (maybe 2 10gigE
> and 2 1gigE?) which one is eth0 supposed to be?   BSD has always used
> driver type in the network device names, and having dealt with device
> confusions before, I understand why.
>
>
>
ethtool can easily tell you the capabilities of the device - you don't need
magic names.

-- 
Stephen Clark
*NetWolves Managed Services, LLC.*
Director of Technology
Phone: 813-579-3200
Fax: 813-882-0209
Email: steve.clark at netwolves.com
http://www.netwolves.com

On Thu, Dec 10, 2015 at 5:32 PM, Matthew Miller <mattdm at mattdm.org>
wrote:
> On Thu, Dec 10, 2015 at 04:56:34PM -0500, m.roth at 5-cent.us wrote:
> > Why not what was suggested, a summary every month or three? How about
> > sending announcements?
>
> Do people _want_ accepted Fedora change announcements posted to this
> list? That's pretty easy to arrange if it really helps. I don't see
a
> big benefit over just following the annoucement list where they're
> posted (filtering out other topics if you want), but if people would
> really find that helpful, we could do it.
>
No.

I'd recommend those who want those announcements subscribe to the proper
Fedora list.
(as you point out, we think alike!)

>
> --
> Matthew Miller
> <mattdm at fedoraproject.org>
> Fedora Project Leader
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> https://lists.centos.org/mailman/listinfo/centos
>


-- 
---~~.~~---
Mike
//  SilverTip257  //