Changes in booting with ubuntu-server 10.04

[Michael Waltz]

After trying out the alpha and beta releases of Ubuntu I am finding that
the booting method of 10.04 is greatly altered. I understand why there
were made, to improve the boot time and make it look nicer, but they
seem to contrast with how a server should boot.

With a traditional unix-like startup a server administrator expects to
see what exactly is booting on the system and if there are any failures
or warnings. It seems that this is not the case in 10.04 so far. When I
first installed Alpha 3, grub automatically booted to the first OS
listing without even listing the menu (I understand you can use ESC to
go into the menu) and then plymouth took over displaying a splash screen
until the login prompt without showing any boot information.

Are there plans for the 10.04 server version to keep with a traditional
server start-up? Or is this the default we should expect upon install?
I've changed a few things, such as enabling the grub menu with a 5
second timeout, removing plymouth on install, but there are still a few
bits that make booting seem odd. The first is that the systems boots by
default to tty7, displaying a blank screen like it's expecting X to
start up. The second is I'm unable to find a way to have upstart display
any information about what init or init.d scripts are running.

There are a few bug reports on these items, but they're not necessarily
"bugs" it seems since this is how the desktop flavor is suppose to work.

With 10.04 server can we expect a traditional server boot or are
individual admins going to have to enable these extra bits of verbosity
manually?

[Egbert Jan van den Bussche]

I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.

Just my 2$c
Egbert Jan (NL)
HCC!Hobbynet admininistrator

-----Oorspronkelijk bericht-----
Verzonden: dinsdag 23 maart 2010 21:57
Onderwerp: Changes in booting with ubuntu-server 10.04
After trying out the alpha and beta releases of Ubuntu I am
finding that
the booting method of 10.04 is greatly altered. I understand
why there
were made, to improve the boot time and make it look nicer, but they
seem to contrast with how a server should boot.
With a traditional unix-like startup a server administrator
expects to
see what exactly is booting on the system and if there are
any failures
or warnings. It seems that this is not the case in 10.04 so
far. When I
first installed Alpha 3, grub automatically booted to the first OS
listing without even listing the menu (I understand you can
use ESC to
go into the menu) and then plymouth took over displaying a
splash screen
until the login prompt without showing any boot information.
Are there plans for the 10.04 server version to keep with a
traditional
server start-up? Or is this the default we should expect upon
install?
I've changed a few things, such as enabling the grub menu with a 5
second timeout, removing plymouth on install, but there are
still a few
bits that make booting seem odd. The first is that the
systems boots by
default to tty7, displaying a blank screen like it's expecting X to
start up. The second is I'm unable to find a way to have
upstart display
any information about what init or init.d scripts are running.
There are a few bug reports on these items, but they're not
necessarily
"bugs" it seems since this is how the desktop flavor is
suppose to work.
With 10.04 server can we expect a traditional server boot or are
individual admins going to have to enable these extra bits of
verbosity
manually?
--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079
--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam

[Fabio T. Leitao]

another peculiar change in the boot, it no longer uses sreadahead, replaced
by ureadahead...

which is just fine by the way, but after I have done a # do-release-upgrade
-d from 9.10 to 10.04 on a test environment, I have found out that the
sreadahead was still installed, together with ureadahead, making the book
impossibly sloooww and full of error in my logs...

might be an easy fix during the upgrade script, and was pretty obvious to
resolve, mainly because it overshut the plymouth GUI all together, echoing
in a bizzar purple terminal (in yellow letters) about the ureadahead exit 4
... pretty ugly combination if you ask me, but I think it was all together
an accident

I should repport in apport as soon as I can recreate the steps...

Post by Egbert Jan van den Bussche
I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.
Just my 2$c
Egbert Jan (NL)
HCC!Hobbynet admininistrator

-----Oorspronkelijk bericht-----
Verzonden: dinsdag 23 maart 2010 21:57
Onderwerp: Changes in booting with ubuntu-server 10.04
After trying out the alpha and beta releases of Ubuntu I am
finding that
the booting method of 10.04 is greatly altered. I understand
why there
were made, to improve the boot time and make it look nicer, but they
seem to contrast with how a server should boot.
With a traditional unix-like startup a server administrator
expects to
see what exactly is booting on the system and if there are
any failures
or warnings. It seems that this is not the case in 10.04 so
far. When I
first installed Alpha 3, grub automatically booted to the first OS
listing without even listing the menu (I understand you can
use ESC to
go into the menu) and then plymouth took over displaying a
splash screen
until the login prompt without showing any boot information.
Are there plans for the 10.04 server version to keep with a
traditional
server start-up? Or is this the default we should expect upon install?
I've changed a few things, such as enabling the grub menu with a 5
second timeout, removing plymouth on install, but there are
still a few
bits that make booting seem odd. The first is that the
systems boots by
default to tty7, displaying a blank screen like it's expecting X to
start up. The second is I'm unable to find a way to have
upstart display
any information about what init or init.d scripts are running.
There are a few bug reports on these items, but they're not
necessarily
"bugs" it seems since this is how the desktop flavor is
suppose to work.
With 10.04 server can we expect a traditional server boot or are
individual admins going to have to enable these extra bits of verbosity
manually?
--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079
--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam

--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam

[Michael Waltz]

I'm getting the same error "ureadahead-other main process (###)
terminated with status 4", although I think it may be due to it running
on ESX. It appears there is a bug report open for it, 484677, with a lot
of replies, but it's marked as invalid.

I'm going to try 10.04 on some physical hardware this week and see if it
gives the same errors.

Micheal

Post by Fabio T. Leitao
another peculiar change in the boot, it no longer uses sreadahead,
replaced by ureadahead...
which is just fine by the way, but after I have done a #
do-release-upgrade -d from 9.10 to 10.04 on a test environment, I have
found out that the sreadahead was still installed, together with
ureadahead, making the book impossibly sloooww and full of error in my
logs...
might be an easy fix during the upgrade script, and was pretty obvious
to resolve, mainly because it overshut the plymouth GUI all together,
echoing in a bizzar purple terminal (in yellow letters) about the
ureadahead exit 4 ... pretty ugly combination if you ask me, but I think
it was all together an accident
I should repport in apport as soon as I can recreate the steps...
I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.
Just my 2$c
Egbert Jan (NL)
HCC!Hobbynet admininistrator

-----Oorspronkelijk bericht-----
Verzonden: dinsdag 23 maart 2010 21:57
Onderwerp: Changes in booting with ubuntu-server 10.04
After trying out the alpha and beta releases of Ubuntu I am finding that
the booting method of 10.04 is greatly altered. I understand why there
were made, to improve the boot time and make it look nicer, but they
seem to contrast with how a server should boot.
With a traditional unix-like startup a server administrator expects to
see what exactly is booting on the system and if there are any failures
or warnings. It seems that this is not the case in 10.04 so far. When I
first installed Alpha 3, grub automatically booted to the first OS
listing without even listing the menu (I understand you can use ESC to
go into the menu) and then plymouth took over displaying a splash screen
until the login prompt without showing any boot information.
Are there plans for the 10.04 server version to keep with a traditional
server start-up? Or is this the default we should expect upon install?
I've changed a few things, such as enabling the grub menu with a 5
second timeout, removing plymouth on install, but there are still a few
bits that make booting seem odd. The first is that the
systems boots by
default to tty7, displaying a blank screen like it's expecting X to
start up. The second is I'm unable to find a way to have upstart display
any information about what init or init.d scripts are running.
There are a few bug reports on these items, but they're not necessarily
"bugs" it seems since this is how the desktop flavor is
suppose to work.
With 10.04 server can we expect a traditional server boot or are
individual admins going to have to enable these extra bits of verbosity
manually?
--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079
--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam

--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam
--
Fábio Leitão
..-. .- -... .. --- .-.. . .. - .- --- ...-.-

--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079

[Fabio T. Leitao]

my test system is a physical hardware (hp dv2000 notebook) but it happens
since the first boot from upgrade.. I have found the same ticket and it oded
me being invalid yet with so many logs... I got to be an issue, though not
in the ureadahead it self, but something solvable

I'm getting the same error "ureadahead-other main process (###) terminated
with status 4", although I think it may be due to it running on ESX. It
appears there is a bug report open for it, 484677, with a lot of replies,
but it's marked as invalid.
I'm going to try 10.04 on some physical hardware this week and see if it
gives the same errors.
Micheal

Post by Fabio T. Leitao
another peculiar change in the boot, it no longer uses sreadahead,
replaced by ureadahead...
which is just fine by the way, but after I have done a #
do-release-upgrade -d from 9.10 to 10.04 on a test environment, I have found
out that the sreadahead was still installed, together with ureadahead,
making the book impossibly sloooww and full of error in my logs...
might be an easy fix during the upgrade script, and was pretty obvious to
resolve, mainly because it overshut the plymouth GUI all together, echoing
in a bizzar purple terminal (in yellow letters) about the ureadahead exit 4
... pretty ugly combination if you ask me, but I think it was all together
an accident
I should repport in apport as soon as I can recreate the steps...
I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.
Just my 2$c
Egbert Jan (NL)
HCC!Hobbynet admininistrator

-----Oorspronkelijk bericht-----
Verzonden: dinsdag 23 maart 2010 21:57
Onderwerp: Changes in booting with ubuntu-server 10.04
After trying out the alpha and beta releases of Ubuntu I am finding that
the booting method of 10.04 is greatly altered. I understand why there
were made, to improve the boot time and make it look nicer, but they
seem to contrast with how a server should boot.
With a traditional unix-like startup a server administrator expects to
see what exactly is booting on the system and if there are any failures
or warnings. It seems that this is not the case in 10.04 so far. When I
first installed Alpha 3, grub automatically booted to the first OS
listing without even listing the menu (I understand you can use ESC to
go into the menu) and then plymouth took over displaying a splash screen
until the login prompt without showing any boot information.
Are there plans for the 10.04 server version to keep with a traditional
server start-up? Or is this the default we should expect upon install?
I've changed a few things, such as enabling the grub menu with a 5
second timeout, removing plymouth on install, but there are still a few
bits that make booting seem odd. The first is that the
systems boots by
default to tty7, displaying a blank screen like it's expecting X to
start up. The second is I'm unable to find a way to have upstart display
any information about what init or init.d scripts are running.
There are a few bug reports on these items, but they're not necessarily
"bugs" it seems since this is how the desktop flavor is
suppose to work.
With 10.04 server can we expect a traditional server boot or are
individual admins going to have to enable these extra bits of verbosity
manually?
--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079
--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam

--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam
--
Fábio Leitão
..-. .- -... .. --- .-.. . .. - .- --- ...-.-

--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079

[Fabio T. Leitao]

By the way, I always try the server distro on a notebook due to the great
chance for problems... when I am happy with it, I try the on a desktop
hardware too... and only then on a server system.. each one has its
particulars, but the OS it self sould be able to work on them all... since
the major difference sould be only the clock ticker (if not mistaken)

the transformation to "desktop" usually is as easy as an apt-get install
ubuntu-desktop anyway (and may be a few autoremoves --purge also)

Post by Fabio T. Leitao
my test system is a physical hardware (hp dv2000 notebook) but it happens
since the first boot from upgrade.. I have found the same ticket and it oded
me being invalid yet with so many logs... I got to be an issue, though not
in the ureadahead it self, but something solvable
I'm getting the same error "ureadahead-other main process (###) terminated

with status 4", although I think it may be due to it running on ESX. It
appears there is a bug report open for it, 484677, with a lot of replies,
but it's marked as invalid.
I'm going to try 10.04 on some physical hardware this week and see if it
gives the same errors.
Micheal

Post by Fabio T. Leitao
another peculiar change in the boot, it no longer uses sreadahead,
replaced by ureadahead...
which is just fine by the way, but after I have done a #
do-release-upgrade -d from 9.10 to 10.04 on a test environment, I have found
out that the sreadahead was still installed, together with ureadahead,
making the book impossibly sloooww and full of error in my logs...
might be an easy fix during the upgrade script, and was pretty obvious to
resolve, mainly because it overshut the plymouth GUI all together, echoing
in a bizzar purple terminal (in yellow letters) about the ureadahead exit 4
... pretty ugly combination if you ask me, but I think it was all together
an accident
I should repport in apport as soon as I can recreate the steps...
I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.
Just my 2$c
Egbert Jan (NL)
HCC!Hobbynet admininistrator

-----Oorspronkelijk bericht-----
Verzonden: dinsdag 23 maart 2010 21:57
Onderwerp: Changes in booting with ubuntu-server 10.04
After trying out the alpha and beta releases of Ubuntu I am finding that
the booting method of 10.04 is greatly altered. I understand why there
were made, to improve the boot time and make it look nicer, but

they

seem to contrast with how a server should boot.
With a traditional unix-like startup a server administrator expects to
see what exactly is booting on the system and if there are any failures
or warnings. It seems that this is not the case in 10.04 so far. When I
first installed Alpha 3, grub automatically booted to the first OS
listing without even listing the menu (I understand you can use ESC to
go into the menu) and then plymouth took over displaying a splash screen
until the login prompt without showing any boot information.
Are there plans for the 10.04 server version to keep with a traditional
server start-up? Or is this the default we should expect upon install?
I've changed a few things, such as enabling the grub menu with a 5
second timeout, removing plymouth on install, but there are still a few
bits that make booting seem odd. The first is that the
systems boots by
default to tty7, displaying a blank screen like it's expecting X to
start up. The second is I'm unable to find a way to have upstart display
any information about what init or init.d scripts are running.
There are a few bug reports on these items, but they're not necessarily
"bugs" it seems since this is how the desktop flavor is suppose to work.
With 10.04 server can we expect a traditional server boot or are
individual admins going to have to enable these extra bits of verbosity
manually?
--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079
--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam

--
ubuntu-server mailing list
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server
More info: https://wiki.ubuntu.com/ServerTeam
--
Fábio Leitão
..-. .- -... .. --- .-.. . .. - .- --- ...-.-

--
Micheal Waltz
SMG Unix Infrastructure
Qualcomm Inc.
Phone: [phone]
Cell: 858-882-7079

--
Fábio Leitão
..-. .- -... .. --- .-.. . .. - .- --- ...-.-

[Etienne Goyer]

Post by Fabio T. Leitao
which is just fine by the way, but after I have done a #
do-release-upgrade -d from 9.10 to 10.04 on a test environment, I have
found out that the sreadahead was still installed, together with
ureadahead, making the book impossibly sloooww and full of error in my
logs...

Sounds like a terrible bug indeed. Is there a bug open on Launchpad
about this one? If not, it would good if you could file one.

[Fabio T. Leitao]

I would say there is, one I have issued, kind of... but since there is no
do-release-upgrade in the lucid branch yet *apport-bug* got a little problem
there... I had to complain about ureadahead in the bug #545596

still unresponded

Post by Fabio T. Leitao
which is just fine by the way, but after I have done a #
do-release-upgrade -d from 9.10 to 10.04 on a test environment, I have found
out that the sreadahead was still installed, together with ureadahead,
making the book impossibly sloooww and full of error in my logs...

Sounds like a terrible bug indeed. Is there a bug open on Launchpad about
this one? If not, it would good if you could file one.
--
Etienne Goyer
Technical Account Manager - Canonical Ltd
Ubuntu Certified Instructor - LPIC-3
~= Ubuntu: Linux for Human Beings =~

[Soren Hansen]

Post by Egbert Jan van den Bussche
I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.

Silly question perhaps, but if they're headless why do you care a whole
lot about what's being shown?

[John Pybus]

Post by Soren Hansen

Post by Egbert Jan van den Bussche
I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.

Silly question perhaps, but if they're headless why do you care a whole
lot about what's being shown?

I'd imagine that it's because, if you've made the effort to head over to
a rack and connect display/keyboard it's not to admire the Ubuntu
branding as you machine starts; something is wrong and you want every
bit of feedback the machine can give you.

Desktop and Server use cases are just diametrically opposite in this
respect. Some of the effort which has gone into smoothing the desktop
boot process can be seen as regression on a server. I guess the effort
on startup time/readahead may be useful on the server, but BIOS startup
times so overwhelmingly dominate ubuntu boot time on my server hardware
that there's very little to be gained even there.

Best,
John

[Michael Waltz]

Just because they're headless doesn't mean that we wouldn't want to see
what all is happening when it boots, especially if there's a problem.

For instance we have a lot of HP and Sun x86 infrastructure, and their
LOMs can display graphics, but we prefer to use the text console for
imaging and troubleshooting. Using a null-modem serial connection is
also a common occurrence on severely distressed hosts. As most of us
sysadmins know 99% of the time a system will boot without an issue, but
it's that 1% of them royally screwing up that we need this type of
verbosity.

Having it boot sequentially and in text (or the options to do so) is
essential for systems a data center. I would think this would be default
and am still perplexed on why a graphical splash screen on boot in the
server edition is default.

I really want Ubuntu Server to flourish and am doing my best to make it
a fully supported as major distribution option in our physical and
virtual data centers. Unfortunately though this effort is stalled due to
not meeting core functionality requirements like this. I've worked
around the issue as much as I can but I've hit a wall since this design
of upstart/plymouth are core features of the 10.04 release.

My boot wish list for Ubuntu Server is this:

1. Grub menu enabled by default with a timeout and "quiet" option
removed in order to display kernel messages on boot.
2. Plymouth removed or disabled, printing all boot messages to tty1 in
standard text with no framebuffer.
3. Upstart in sysvinit compatibility mode, booting init scripts in order
and not relying on event triggers.

I don't want it to seem like I'm making demands and am more than willing
to help out. I have access to datacenter grade hardware/software (HP and
Sun x86 blade centers and single systems, and ESX, XEN, KVM virtual
clusters) and am willing to test and develop for these platforms.

Micheal

Post by Soren Hansen

Post by Egbert Jan van den Bussche
I have not tried any of the alpha or beta releases yet. But now I see
Micheal's horror story. PLEASE do not assume that a server can display ANY
kind of graphics. All my servers run headles and can only be logged in with
ssh. If really no other way, I have to bring a keyboard/display to the
rack...
Again please let the init messages flow over our consoles.

Silly question perhaps, but if they're headless why do you care a whole
lot about what's being shown?

[Egbert Jan]

-----Oorspronkelijk bericht-----
Verzonden: donderdag 25 maart 2010 19:22
Aan: Egbert Jan van den Bussche
Onderwerp: Re: Changes in booting with ubuntu-server 10.04
On Tue, Mar 23, 2010 at 10:07:51PM +0100, Egbert Jan van den

Post by Egbert Jan van den Bussche
I have not tried any of the alpha or beta releases yet. But

now I see

Post by Egbert Jan van den Bussche
Micheal's horror story. PLEASE do not assume that a server

can display

Post by Egbert Jan van den Bussche
ANY kind of graphics. All my servers run headles and can only be
logged in with ssh. If really no other way, I have to bring a
keyboard/display to the rack... Again please let the init messages
flow over our consoles.

Silly question perhaps, but if they're headless why do you
care a whole lot about what's being shown?
--
Soren Hansen
Ubuntu Developer
http://www.ubuntu.com/

Well, it is always possible to hook something up. Not nessarily capable of
doing fancy graphics.

Egbert Jan

[Egbert Jan]

Hi again Soren,

Because IF I need to attach a console I do that to catch each and every
information that could help me find any problems during boot.

I'm from a old era. VAXVMS on DEC mini's. There were only servers and
character terminals. There were no VGA output devices. I want (and I think
many with me) the simple all revealing line oriented logging during the boot
process on a simple serial or console port. No bells, no wistles. If
something goes wrong during the boot we want to know and have informative
output at hand on (simple) output devices. Not hidden in high level messages
in fancy screens.

But what heck, nobody asked to have fancy server bootspash screens on
servers. If that is the new way of working, fine but just give us the
possibility to have a boot option to have all starting services scrolling
over the screen (and hopefully an 'OK' on the right).

No hard feelings, I also want this distro to be the best.
Egbert Jan (NL)

PS I switched from years and years of Mandriva usage to Ubuntu because I got
the impression that Ubuntu developers WERE paying attention to what the
community wants.

-----Oorspronkelijk bericht-----
Verzonden: donderdag 25 maart 2010 23:02
Aan: Egbert Jan
Onderwerp: Re: Changes in booting with ubuntu-server 10.04

I just expressed my fear that the Good Old boot messages are being
replaced by nothing-saying desktop output. IF

ubuntu/canonical really

keeps this new booting as default with no easy install

option, I'm off

to Debian, but with bleeding heart.

I still don't understand your argument. Why do you care
what's on the VGA output (or not) if you don't have a monitor
attached to it?
--
Soren Hansen
Ubuntu Developer
http://www.ubuntu.com/

[carlopmart]

Post by Egbert Jan
Hi again Soren,
Because IF I need to attach a console I do that to catch each and every
information that could help me find any problems during boot.
I'm from a old era. VAXVMS on DEC mini's. There were only servers and
character terminals. There were no VGA output devices. I want (and I think
many with me) the simple all revealing line oriented logging during the boot
process on a simple serial or console port. No bells, no wistles. If
something goes wrong during the boot we want to know and have informative
output at hand on (simple) output devices. Not hidden in high level messages
in fancy screens.
But what heck, nobody asked to have fancy server bootspash screens on
servers. If that is the new way of working, fine but just give us the
possibility to have a boot option to have all starting services scrolling
over the screen (and hopefully an 'OK' on the right).
No hard feelings, I also want this distro to be the best.
Egbert Jan (NL)
PS I switched from years and years of Mandriva usage to Ubuntu because I got
the impression that Ubuntu developers WERE paying attention to what the
community wants.

+1 and totally agree with Egbert.

If Ubuntu/Canonical would like to have a serious server ditro, usplash, plymouth and
similiar technologies aren't need on server edition and they were disabled by default.

Server edition needs to be stable, robust, administrable via command line. The
remaining surplus.

[Thierry Carrez]

Post by Egbert Jan
But what heck, nobody asked to have fancy server bootspash screens on
servers. If that is the new way of working, fine but just give us the
possibility to have a boot option to have all starting services scrolling
over the screen (and hopefully an 'OK' on the right).

Please follow status on Launchpad bug 548954:
https://bugs.launchpad.net/ubuntu/+source/ubuntu-meta/+bug/548954

That bug doesn't need comments on how desirable it is, though patches
and testers are always welcome.

['Soren Hansen']

Post by Egbert Jan
Because IF I need to attach a console I do that to catch each and
every information that could help me find any problems during boot.

The boot process as it is (and has been for years) already doesn't give
a whole lot of information. You usually have correlate its output with
syslog and/or dmesg and/or deamon specific log files to really narrow
down on a problem.

That said, Plymouth is actually supposed to make this /better/. I'm not
sure if that's going to happen for Lucid (it's not really my area), but
the idea is that since there's something in charge of collecting output
from boot scripts and presenting it to the user, that something can also
put this information in a log file. This means that that anyhing you see
during boot should end up in a log file which should alleviate the need
for looking at the boot sequence.

Yes, that's a lot of "should"'s, but I'm afraid that's all I have right
now. Someone else may be able to weigh in with some more authoritative
information or at least more detail.

Post by Egbert Jan
But what heck, nobody asked to have fancy server bootspash screens on
servers.

That's simply not true. /You/ may not have asked for it, but it's
certainly been asked for. I myself, for instance, don't mind a pretty
boot sequence (brief as it may be).

Post by Egbert Jan
PS I switched from years and years of Mandriva usage to Ubuntu because
I got the impression that Ubuntu developers WERE paying attention to
what the community wants.

I like to think that we do. However, please consider that the community
is diverse as are its opinions on different matters.

[Etienne Goyer]

Post by 'Soren Hansen'

Post by Egbert Jan
But what heck, nobody asked to have fancy server bootspash screens on
servers.

That's simply not true. /You/ may not have asked for it, but it's
certainly been asked for. I myself, for instance, don't mind a pretty
boot sequence (brief as it may be).

I have not seen anybody complaining on the look of the Server Edition
boot process either. Was that discussed at a UDS, or something? If so,
I must have missed the blueprint.

Just because of the potential for regressions and unforeseen problems, I
think it is a terrible idea to introduce that feature in an LTS cycle.
I hope it get backed out before release.

[Alvin]

Post by Etienne Goyer

Post by 'Soren Hansen'

Post by Egbert Jan
But what heck, nobody asked to have fancy server bootspash screens on
servers.

That's simply not true. /You/ may not have asked for it, but it's
certainly been asked for. I myself, for instance, don't mind a pretty
boot sequence (brief as it may be).

I have not seen anybody complaining on the look of the Server Edition
boot process either. Was that discussed at a UDS, or something? If so,
I must have missed the blueprint.
Just because of the potential for regressions and unforeseen problems, I
think it is a terrible idea to introduce that feature in an LTS cycle.
I hope it get backed out before release.

I haven't seen the Lucid boot process yet, and that is the sole reason for not
having complained yet. Currently, we lack every form of boot logging. Some
bugs on Launchpad have pictures of the boot process attached to them, taken by
a digital camera.

In Karmic, there were several bugs introduced by mountall/upstart/plymouth.
See the latest comments of https://bugs.launchpad.net/bugs/470776
Error messages will be hidden by default.
The specific error messages here can sometimes be correct, and sometimes be
wrong. Most people will rather see the source of that problem fixed instead of
covered up with a nice animation.

As a server administrator, I'm not interested in fast boot times, nor in fancy
graphics. I'm interested in reliable booting and knowing what is going on. I'd
like to have upstart because it eliminates the need to set a specific order in
the processes (no more rcX), but I'd never sacrifice reliability for that.

[Nikolai K. Bochev]

Didn't centos have a boot splash that can show the boot process in a small console window ? Or was it SuSe ?
So you get the pretty booting splash with a nice progress bar, but if you want, you can click on an arrow to open the console and watch the boot process.
It's both "enterprisey" and informative.

Post by 'Soren Hansen'
The boot process as it is (and has been for years) already doesn't give
a whole lot of information. You usually have correlate its output with
syslog and/or dmesg and/or deamon specific log files to really narrow
down on a problem.
That said, Plymouth is actually supposed to make this /better/. I'm not
sure if that's going to happen for Lucid (it's not really my area), but
the idea is that since there's something in charge of collecting output
from boot scripts and presenting it to the user, that something can also
put this information in a log file. This means that that anyhing you see
during boot should end up in a log file which should alleviate the need
for looking at the boot sequence.
Yes, that's a lot of "should"'s, but I'm afraid that's all I have right
now. Someone else may be able to weigh in with some more authoritative
information or at least more detail.
That's simply not true. /You/ may not have asked for it, but it's
certainly been asked for. I myself, for instance, don't mind a pretty
boot sequence (brief as it may be).
I like to think that we do. However, please consider that the
community
is diverse as are its opinions on different matters.

[Tom Ellis]

Post by Nikolai K. Bochev
Didn't centos have a boot splash that can show the boot process in a small console window ? Or was it SuSe ?
So you get the pretty booting splash with a nice progress bar, but if you want, you can click on an arrow to open the console and watch the boot process.
It's both "enterprisey" and informative.

This was RHGB which has it's own bunch of issues, which Plymouth was
designed to replace. Plymouth is the way forward, but I agree we need to
be able to display some output of the standard init scripts like in
previous releases. Removing splash and quiet increases verbosity of the
boot sequence but I still see zero output from the run of the mill
server services, apache etc.

This may be just a bug with where the output is being redirected,
perhaps we can leverage some of the work in other distributions running
Plymouth to see if the same issue is encountered?

- --
Regards,
Tom Ellis
Premium Service Engineer - Canonical
GPG: EEC4 4552 B57E D9BD 7E57 F7F5 3990 6F7D 063C 355A

[Imre Gergely]

Post by Tom Ellis

Post by Nikolai K. Bochev
Didn't centos have a boot splash that can show the boot process in a small console window ? Or was it SuSe ?
So you get the pretty booting splash with a nice progress bar, but if you want, you can click on an arrow to open the console and watch the boot process.
It's both "enterprisey" and informative.

This was RHGB which has it's own bunch of issues, which Plymouth was
designed to replace. Plymouth is the way forward, but I agree we need to
be able to display some output of the standard init scripts like in
previous releases. Removing splash and quiet increases verbosity of the
boot sequence but I still see zero output from the run of the mill
server services, apache etc.
This may be just a bug with where the output is being redirected,
perhaps we can leverage some of the work in other distributions running
Plymouth to see if the same issue is encountered?

It's not really a server issue, but it could be related to booting. I've
just updated my Karmic to Lucid on my laptop. I have an ecrypted home
partition (with LUKS). I did not install it this way, but converted it
later my hand.

Now with Lucid I have problems on every boot. It asks me for the
password (a nice textbox appears on the splash screen), but then it just
waits there and I have to press S (skip) or M (mount/manual?), which
doesn't work and I get a shell where I can mount it manually.

I tried with nofb and without quiet and splash at the beginning, now it
asks me in text mode, but the same thing happens, it's just sitting
there, after I input the passphrase. I'm not even mentioning that
there's no clear prompt because it's overwritten with other messages /
scrolls up.

Does anybody else have a problem with encrypted partitions at boot time?