I should add that I have no experience of RHEL 7 as of yet. It appears
they're doing something new/different with the naming schemes of
interfaces, and I can't say if it's for the better.
I've used BIOSDEVNAME=1 on all my clusters from CentOS 6.3 to 6.5. I
like knowing that em1 is really my first embedded nic. I like knowing
that "p3p2" is the second HCA on the Mellanox card in my third PCI slot.
That's great for me.
I will say that I've run into cases where BIOSDEVNAME gets confused. It
seems to work best on Dell hardware -- I think Dell contributed the
original code. On some IBM hardware, it interprets the onboard SATA/SAS
controller as "em1", making your first embedded NIC "em2". That is
really unfortunate. But for me that's the only downside I've found,
yet, of using the BIOSDEVNAME approach.
Perhaps changes in RHEL 7 are meant to prevent mistakes like the one
I've just mentioned.
On 02/11/2014 11:49 AM, Jarrod Johnson wrote:
> Yes, that's the tricky thing for me. It does little but help the most
> trivial case ('eth0' becomes 'em1', without resorting to highly fragile
> persistant name schemes, and with a stronger guarantee that it is the
> port you intuitively expect). There is a middle ground where it can go
> bonkers (eth1 becoming enp12s0f1, usb0 becoming enp0s29u1u1u5 and it's
> much harder to know what the device will be named). Then of course it
> comes back around with the knowledge that at least enp12s0f1 in theory
> should be a consistent name for a replacement device. For my part, I've
> been endeavoring to make xCAT not really care at all (and as a testament
> to that, I didn't even notice the device changes until after the
> deployment was completed and that 'em1' was showing the configuration
> instead of 'eth0'). I frankly don't have much of a horse in this race.
> My inclination is to let it ride and document bootparams.addkcmdline
> for people that don't want it. For one, if there is an opt-in/opt-out,
> it's much easier if the semantics match the distro, opt-out through that
> setting versus some new invented syntax to make it removed. For
> another, I'm strongly considering giving xCAT the ability to curate
> specific names for discovered nics, and it's all the easier if people
> tend to want to give the xCAT curated nics names like 'eth0' and all the
> other disused nics are helpfully moved out of the way by being named
> weird and goofy names.
>
> Anyway, that's my current inclination after giving it some thought. The
> biggest thoughts I have on administering RHEL7:
> -I'm glad I switched to iproute2 a long time ago. The new, optional
> ifconfig would have made me redo a fair bit of scripting
> -Finally, 'alt-f2' like capability during install for pure rcons usage
> (they put the text installer in tmux, with the only downside I can think
> of being ANSI control sequences in console log).
> -I haven't made up my mind about systemd yet.....
>
>
> On Mon, Feb 10, 2014 at 7:25 PM, Jonathan Mills <[email protected]
> <mailto:[email protected]>> wrote:
>
> I've gone the other way, fully embracing BIOSDEVNAME=1. For me, it
> makes certain things a whole lot easier.
>
> > On Feb 10, 2014, at 6:55 PM, "Fenn, Michael"
> <[email protected]> wrote:
> >
> > biosdevname=0 for life. :)
> >
> > Michael
> >
> >> On 2/10/14 6:08 PM, "David Johnson" <[email protected]
> <mailto:[email protected]>> wrote:
> >>
> >> I'm still struggling to keep up with the changes that rhel6
> imposed on
> >> us. If xcat can go from MAC address to ipv4 address and make
> the bits
> >> come out of the right wire, I'll be happy. Half my nodes come up
> with
> >> eth0-remamed
> >>
> >> -- ddj
> >> Dave Johnson
> >>
> >>> On Feb 10, 2014, at 5:48 PM, Jarrod Johnson
> >>> <[email protected]
> <mailto:[email protected]>> wrote:
> >>>
> >>> So if we apply the typical amount of conservative discipline to the
> >>> package list, at least the following commands will not be
> installed:
> >>> ifconfig
> >>> lspci
> >>> lsusb
> >>>
> >>> ifconfig output has changed significantly to boot if we do install.
> >>>
> >>> Should we include these packages or are people comfortable with
> >>> alternative ways into the data?
> >>>
> >>> Also, should we slap down the nic renaming from 'eth1' to
> 'enp12s0f1'
> >>> or leave that intact or explicitly name the nics that we care
> about and
> >>> let the other nics get named whatever?
> >>>
> >>>
> -------------------------------------------------------------------------
> >>> -----
> >>> Android apps run on BlackBerry 10
> >>> Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
> >>> Now with support for Jelly Bean, Bluetooth, Mapview and more.
> >>> Get your Android app in front of a whole new audience. Start now.
> >>>
> >>>
> http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clk
> >>> trk
> >>> _______________________________________________
> >>> xCAT-user mailing list
> >>> [email protected]
> <mailto:[email protected]>
> >>> https://lists.sourceforge.net/lists/listinfo/xcat-user
> >>
> >>
> --------------------------------------------------------------------------
> >> ----
> >> Android apps run on BlackBerry 10
> >> Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
> >> Now with support for Jelly Bean, Bluetooth, Mapview and more.
> >> Get your Android app in front of a whole new audience. Start now.
> >>
> http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clkt
> >> rk
> >> _______________________________________________
> >> xCAT-user mailing list
> >> [email protected]
> <mailto:[email protected]>
> >> https://lists.sourceforge.net/lists/listinfo/xcat-user
> >
> >
> >
>
> ------------------------------------------------------------------------------
> > Android apps run on BlackBerry 10
> > Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
> > Now with support for Jelly Bean, Bluetooth, Mapview and more.
> > Get your Android app in front of a whole new audience. Start now.
> >
>
> http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clktrk
> > _______________________________________________
> > xCAT-user mailing list
> > [email protected]
> <mailto:[email protected]>
> > https://lists.sourceforge.net/lists/listinfo/xcat-user
>
>
> ------------------------------------------------------------------------------
> Android apps run on BlackBerry 10
> Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
> Now with support for Jelly Bean, Bluetooth, Mapview and more.
> Get your Android app in front of a whole new audience. Start now.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clktrk
> _______________________________________________
> xCAT-user mailing list
> [email protected] <mailto:[email protected]>
> https://lists.sourceforge.net/lists/listinfo/xcat-user
>
>
--
Jonathan Mills
Systems Administrator
Renaissance Computing Institute
UNC-Chapel Hill
------------------------------------------------------------------------------
Android apps run on BlackBerry 10
Introducing the new BlackBerry 10.2.1 Runtime for Android apps.
Now with support for Jelly Bean, Bluetooth, Mapview and more.
Get your Android app in front of a whole new audience. Start now.
http://pubads.g.doubleclick.net/gampad/clk?id=124407151&iu=/4140/ostg.clktrk
_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user
