client: searching for jumpstart directory: not found. No entry in rules file.
bertieb...@gmail.com
/etc/bootparams seems to be doing its job because client boots up ok
and seems to find the correct sysidcfg.
However, for some reason its saying rules is wrong.
What to check first?
hume.sp...@bofh.ca
> /etc/bootparams seems to be doing its job because client boots up ok
> and seems to find the correct sysidcfg.
>
> However, for some reason its saying rules is wrong.
The error (which you put in the subject line, making it very hard to read)
seems to indicate rather that it can't find the jumpstart dir. I would
interpret that to mean it can't find the rules file at all.
> What to check first?
Is this a SPARC or x86 system? Are you using RARP or DHCP? And what do
you have specified for the install_config parameter?
--
Brandon Hume - hume -> BOFH.Ca, http://WWW.BOFH.Ca/
Richard B. Gilbert
Your rules file?
bertieb...@gmail.com
wrote:
BTW. Should mention it is there....
Richard B. Gilbert
The documentation at this link:
http://docs.sun.com/app/docs/doc/821-0439
might be of some help.
Richard B. Gilbert
Is are the contents syntactically and semantically correct?
I dimly recall that there are two files involved: sysidconfig is one and
the other is your rules file.
The "Advanced Installation Manual" goes into all this in some detail
bertieb...@gmail.com
Thing is I saved a few of the files and the complete /jumpstart
directory off another working js server.
It seems to find the correct profile and then fail to find the
jumpstart directory? Weird - because I think I'm right in saying this
is all identified in the bootparams file, isnt it?
Richard B. Gilbert
Don't know! The one time in my career that it would have been useful, I
hadn't a clue. Somebody showed me how to install from the local CD
drive. I repeated the steps for eight machines, one at a time. Since
then I've read TFM but had no need to do mass installations.
The best advice I can give you is RTFM and follow the directions.
Cydrome Leader
jumpstart is picky and overly complex monster. On x86, it's just
worthless.
always add the -v flag to your boot install whatever command so you can
see what's happening and where it blows up.
You're at least able to get an address, and boot the kernel off your
jumpstart machine if you're getting rules errors- this is good.
Richard B. Gilbert
If Jumpstart fails to work as documented, you need to complain to Sun.
I don't know how seriously they take reports/complaints from people who
are not paying customers. I'm fairly sure that they will fix it
someday. I couldn't even attempt to guess what priority the problem
would be given unless the complaint came from a customer paying support.
OTOH, Jumpstart is not particularly useful unless you are installing two
or more machines. The more machines you have to install the more
valuable it is.
If you are installing one system just put the DVD in the drive and boot
from it! Be ready to supply the IP address, netmask, node name, and
timezone for the machine you are installing. You will also be asked for
the current date and time. There is nothing particularly arcane about
the installation. It's mostly waiting for the next time your input is
required.
You will be asked how to partition the disk. Allow at least 4GB for /
plus /usr. Allow another 4GB for /var. If you have reason to think you
will need more, go ahead and allocate more. Disk space is relatively
cheap.
You will also be asked if you want to install everything or not. Unless
you are VERY certain that you do not want to install some part of
Solaris just go ahead and install everything. If you don't you will
probably regret it later. Adding pieces of the O/S later requires that
you know what all the dependencies are. While you can figure it out,
you'll wish you had done the full install.
YTC#1
If you don't want to spend hours delving around trying to sort it out, I
suggest you go and download JET, JumpStart Enterprise Toolkit, from
http://www.sun.com/downloads.
Have a read of
http://wikis.sun.com/display/JET/Home
The authors' (Mike Ramchand) blog at http://blogs.sun.com/mramcha/
Quick start guide at
http://www.sun.com/bigadmin/content/jet/
External Sun help at
http://jet.maui.co.uk/wiki/index.php/Main_Page
And the Yahoo Group at JETJum...@yahoogroups.com
I know JumpStart can be finicky etc, hence why JET is such a good tool.
Personally I have not istalled by DVD for years, even a single box, I
just plug my laptop in and JumpStart from there.
Hell, I even used JET to JumpStart 2 RedHat servers last week.
--
Bruce Porter
"The internet is a huge and diverse community and not every one is friendly"
http://www.ytc1.co.uk
There *is* an alternative! http://www.openoffice.org/
Heinz M�ller
"BertieB...@gmail.com" <bertieb...@gmail.com> schrieb im Newsbeitrag
news:906c2c3f-802d-43f8...@a32g2000yqm.googlegroups.com...
Check if the share is set correct(ly) on the jumpstartserver.
Heinz
Cydrome Leader
they don't care if paying customers complain. they don't even care if
sales reps trying to peddle new hardware get complaints during
trial runs. Then of course, sun fired off their sales people, so it's
not important anyways anymore.
> someday. I couldn't even attempt to guess what priority the problem
> would be given unless the complaint came from a customer paying support.
been there done, that, I have a long list of support contracts here.
need a list of cases about jump start on x86 being a total piece of shit?
Richard B. Gilbert
Nope! I run Solaris on Sparc workstations.
Cydrome Leader
It's all sparc/niagara whatever here. We tried the x86 stuff, it was complete
garbage- hardware and software, plus third party "business" apps for solaris
x86 is pretty much zero.
YTC#1
Which bits in particular are causing trouble ?
Cydrome Leader
By using a non-sun DHCP server, we finally got a kernel to boot. rules
never worked, ever. Nobody at sun could get it to work either, which
doesn't surprise me as they keep changing all the secret paramaters and
settings every 9 seconds with their junior attempts as PXE booting.
To this day, there's still no accurate or even close to upto date doc on
jumpstarting sun stuff with PXE where the commands as they show actually
work. There's some "blueprint" from 2006 or so. it's completely useless.
Sun's attempt at remote console for their own x86 hardware is a complete
joke as well.
wan booting sparc machines with newer firmware works better, but only if
you use JET to make the 6000 character long commands.
If anybody can explain why a MAC address and hostname have to be used at
least 50 times per client, I'd love to hear the logic behind that.
Andrew Deason
I can assure you this has worked for me:
<http://www.acm.uiuc.edu/admin/wiki/Jumpstart>. That is incomplete as it
doesn't document the framework for begin/finish scripts, but you
shouldn't care about that unless you want to use our begin/finish script
framework.
--
Andrew Deason
adea...@uiuc.edu
YTC#1
Which is an outright slur, no one you asked at Sun gave you the right
anser, as I keep pointing out JET (which is also the engine behinbd N1 and
XvM JumpStarts can do all you requested so far, including using non Sun
DHCP. And it was written and supported by Sun staff.
> doesn't surprise me
as they keep changing all the secret paramaters and
> settings every 9 seconds with their junior attempts as PXE booting.
PXE booting is a bit of a nightmare, very few seem to understand it well.
Mike Ramchand has done a lot of work in that area to simplyfy the usage
for JumpStart.
> To this day, there's still no accurate or even close to upto date doc on
> jumpstarting sun stuff with PXE where the commands as they show actually
> work. There's some "blueprint" from 2006 or so. it's completely useless.
Ok, I'll go for that one, but bear inb mind that PXE booting isnot Sun
technology.
>
> Sun's attempt at remote console for their own x86 hardware is a complete
> joke as well.
Depends which one you use :-)
But as I always prefer to stay away from GUIs I seem to have less
problems, and the emote console has nothing to do with JumpStart.
>
> wan booting sparc machines with newer firmware works better, but only if
> you use JET to make the 6000 character long commands.
Aha, so you have played with it then :-)
By and large, WAN boot is a PITA, and not that many *really* need it, a
combo of DHCP/NFS often works better.
>
> If anybody can explain why a MAC address and hostname have to be used at
> least 50 times per client, I'd love to hear the logic behind that.
Again, not sure what your problem is here, you supply it *once* to a
JumpStart tool like JET. The tool then handles how and when they get
handed to the client.
I think 50 is an over exaggeration.
Cydrome Leader
I should not have to use a side project tool to jumpstart a machine. If I
do, the process is just garbage.
Jet isn't a flawless tool either by the way.
>> doesn't surprise me
> as they keep changing all the secret paramaters and
>> settings every 9 seconds with their junior attempts as PXE booting.
>
> PXE booting is a bit of a nightmare, very few seem to understand it well.
> Mike Ramchand has done a lot of work in that area to simplyfy the usage
> for JumpStart.
PXE booting in ONLY a nightmare when sun is involved.
microsoft has had a RIS service forever, it works perfectly. It doesn't
even require registry hacks to get working.
you hit f12 (or whatever) to netbook your client, possibly login to a
small DOS looking program that loads, and come back to a fully installed
workstation.
go bother some IT guy to see how it works, and has worked for like 10
years now.
It's really easy stuff. You don't have to add console flags to grub menus
and stupid shit like that.
linux people are mostly retarded, but they sure like reinstalling their
machines every 15 minutes and have PXE booting all figured out too.
Last time I looked, the FreeBSD people just didn't get it either. but
their entire project is becoming increasingly obsolete anyways which is
too bad.
>> To this day, there's still no accurate or even close to upto date doc on
>> jumpstarting sun stuff with PXE where the commands as they show actually
>> work. There's some "blueprint" from 2006 or so. it's completely useless.
>
>
> Ok, I'll go for that one, but bear inb mind that PXE booting isnot Sun
> technology.
Sun has no problem making and touting intel hardware, so mayb ethey need
to attend some seminar at intel HQ about PXE.
>> Sun's attempt at remote console for their own x86 hardware is a complete
>> joke as well.
>
>
> Depends which one you use :-)
> But as I always prefer to stay away from GUIs I seem to have less
> problems, and the emote console has nothing to do with JumpStart.
Sun remote consoles on their PC harware is completely bogus. If somebody
leaves a monitor cable connected to the server, you can't grab the
non-serial consoles remotely, vs. HP's ILO where both people can type and
see the same thing at the same time as their remote console REALLY is the
same as the hardare one.
>> wan booting sparc machines with newer firmware works better, but only if
>> you use JET to make the 6000 character long commands.
>
> Aha, so you have played with it then :-)
>
> By and large, WAN boot is a PITA, and not that many *really* need it, a
> combo of DHCP/NFS often works better.
Anything works better than obsolete nonsense like RAPR servers and
symlinks in some tftpboot directory with lists of mac addresses.
Any time I have to deal with mac addresses, I feel like it's early 1990s
and I'm installing the Novell client on some PC running DOS, or I'm
configuring a JetLan print server for some "letter quality" dot matrix
printer.
You also mentioned you prefer CLI stuff over a GUI, so that rules out
using the sun DHCP server. That thing is a strange hunk of junk. MS and
ISC did a far better job. Sun should just drop their attempt and move on.
Oh wait, maybe a team of bored sun people in the UK office wrote a front
end to the command line for the sun DHCP server that inputs and outputs
sensible values because the people who wrote the DHCP server never
bothered to do it themselves.
>> If anybody can explain why a MAC address and hostname have to be used at
>> least 50 times per client, I'd love to hear the logic behind that.
>
> Again, not sure what your problem is here, you supply it *once* to a
> JumpStart tool like JET. The tool then handles how and when they get
> handed to the client.
>
> I think 50 is an over exaggeration.
ok how about 35 times?
To any nay-sayers, just watch how you net install windows.
phil.go...@bolthole.com
> By using a non-sun DHCP server, we finally got a kernel to boot. rules
> never worked, ever. Nobody at sun could get it to work either, which
> doesn't surprise me as they keep changing all the secret paramaters and
> settings every 9 seconds with their junior attempts as PXE booting.
>
> To this day, there's still no accurate or even close to upto date doc on
> jumpstarting sun stuff with PXE where the commands as they show actually
> work. There's some "blueprint" from 2006 or so. it's completely useless.
FYI: I had some problems in that area myself.
I discovered by a lot of trial and error, that
#1. it doesnt like subdirectories for profiles. has to be in same dir
as rules file.
(Same problem with using WANBOOT on sparc now!!!)
#2. I gave up on regular rules/profile sharing for PXE/sol x86,and
went to a little known method of using a tarfile.
My grub entries look like the following:
title 6 Solaris 10 net install experimental tarcfg, serial console
kernel /boot/200810/multiboot kernel/unix - install nfs://x.x.x.x
/jumpstart/jumpstart.tar -B console=ttya,install_media=nfs://x.x.x.x/
jumpstart/s10x86200810
module /boot/200810/x86.miniroot
Even WANBOOT, ugly as it it, would be an improvement over this mess.
Dammit sun, pick a technology and STANDARDIZE!!
Tips:
"/boot" entries, are references to a subdir in /tftpboot
jumpstart.tar is created by a Makefile in the jumpstart dir that looks
like the following:
FILES = README rules rules.ok sysidcfg *.prof begin_* finish_*
jumpstart.tar: $(FILES)
tar cvf jumpstart.tar $(FILES)
YTC#1
I would not term it a side project, as it is in N1/XVM
>
> Jet isn't a flawless tool either by the way.
Name me a flawless tool
>
>>> doesn't surprise me
>> as they keep changing all the secret paramaters and
>>> settings every 9 seconds with their junior attempts as PXE booting.
>>
>> PXE booting is a bit of a nightmare, very few seem to understand it well.
>> Mike Ramchand has done a lot of work in that area to simplyfy the usage
>> for JumpStart.
>
> PXE booting in ONLY a nightmare when sun is involved.
Disagree.
>
> microsoft has had a RIS service forever, it works perfectly. It doesn't
> even require registry hacks to get working.
Can't comment, not done it. However, can RIS be used to install Solaris ?
I've reckon JET could be amended to install Windows (using RIS)
>
> you hit f12 (or whatever) to netbook your client, possibly login to a
> small DOS looking program that loads, and come back to a fully installed
> workstation.
I hit F12 on a server , and it just goes and installs my client as per
previous instructions. Ok, so I had to tie it to a MAC address, because
not all machines are the same. I have seen some fancy hacks where someone
has got PXE to halt, select a template to continue, but that *is not*
hands free.
>
> go bother some IT guy to see how it works, and has worked for like 10
> years now.
Grumpy, arn't you ?
Strange, I only normally use the SUN dhcp server, must be my imagination
that I do that via the CLI.
>
> Oh wait, maybe a team of bored sun people in the UK office wrote a front
From experience, they have no time to be bored, they do stuff in their own
time
> end to the command line for the sun DHCP server that inputs and
outputs
> sensible values because the people who wrote the DHCP server never
> bothered to do it themselves.
>
>>> If anybody can explain why a MAC address and hostname have to be used
>>> at least 50 times per client, I'd love to hear the logic behind that.
>>
>> Again, not sure what your problem is here, you supply it *once* to a
>> JumpStart tool like JET. The tool then handles how and when they get
>> handed to the client.
>>
>> I think 50 is an over exaggeration.
>
> ok how about 35 times?
Sorry, bit from my POV, its once.
What happens under the bonnet is not a concern.
>
> To any nay-sayers, just watch how you net install windows.
I'm still watching, nothing happening
Cydrome Leader
> On Dec 23, 1:52?pm, Cydrome Leader <prese...@MUNGEpanix.com> wrote:
>
>> By using a non-sun DHCP server, we finally got a kernel to boot. rules
>> never worked, ever. Nobody at sun could get it to work either, which
>> doesn't surprise me as they keep changing all the secret paramaters and
>> settings every 9 seconds with their junior attempts as PXE booting.
>>
>> To this day, there's still no accurate or even close to upto date doc on
>> jumpstarting sun stuff with PXE where the commands as they show actually
>> work. There's some "blueprint" from 2006 or so. it's completely useless.
>
> FYI: I had some problems in that area myself.
> I discovered by a lot of trial and error, that
>
> #1. it doesnt like subdirectories for profiles. has to be in same dir
> as rules file.
> (Same problem with using WANBOOT on sparc now!!!)
nice. I wonder how you found this, or how sun forgot the cd command.
> #2. I gave up on regular rules/profile sharing for PXE/sol x86,and
> went to a little known method of using a tarfile.
I checked my notes and case history and found this is what wasn't working.
No matter what the rules were, no host would ever match a profile and the
install would be interactive. Any rules we made or recycled for sparc
worked fine though, so there wasn't a clueless factor. This wasn't a case
of no nfsv domain- a client being jumped would never match anything except
the generic profile for an i86 client. This was solaris 10 08/06.
> My grub entries look like the following:
>
> title 6 Solaris 10 net install experimental tarcfg, serial console
> kernel /boot/200810/multiboot kernel/unix - install nfs://x.x.x.x
> /jumpstart/jumpstart.tar -B console=ttya,install_media=nfs://x.x.x.x/
> jumpstart/s10x86200810
> module /boot/200810/x86.miniroot
>
> Even WANBOOT, ugly as it it, would be an improvement over this mess.
> Dammit sun, pick a technology and STANDARDIZE!!
they can't even pick names for established products. I noticed in meetings
everybody knows what "apache" is, but depending on how long the person has
been around the sun webserver is "SJWS", "iPlanet" or just a generic and
safe "sun webserver".
> Tips:
>
> "/boot" entries, are references to a subdir in /tftpboot
>
> jumpstart.tar is created by a Makefile in the jumpstart dir that looks
> like the following:
>
>
> FILES = README rules rules.ok sysidcfg *.prof begin_* finish_*
>
> jumpstart.tar: $(FILES)
> tar cvf jumpstart.tar $(FILES)
weird, I'll need to try this. I wonder why they even made it, or used
something easy like tar. I'd expect it to need cpio-ed file like a flar or
something you can't open and examine again.
Cydrome Leader
I have no idea what N1/XVM is, or XPG-3000 on MegaStar-321 rev 45
>> Jet isn't a flawless tool either by the way.
>
> Name me a flawless tool
halt has never let me down. init and reboot different stories.
>>>> doesn't surprise me
>>> as they keep changing all the secret paramaters and
>>>> settings every 9 seconds with their junior attempts as PXE booting.
>>>
>>> PXE booting is a bit of a nightmare, very few seem to understand it well.
>>> Mike Ramchand has done a lot of work in that area to simplyfy the usage
>>> for JumpStart.
>>
>> PXE booting in ONLY a nightmare when sun is involved.
>
> Disagree.
share some examples.
>> microsoft has had a RIS service forever, it works perfectly. It doesn't
>> even require registry hacks to get working.
>
> Can't comment, not done it. However, can RIS be used to install Solaris ?
> I've reckon JET could be amended to install Windows (using RIS)
If you like reinventing the wheel, go for it.
>> you hit f12 (or whatever) to netbook your client, possibly login to a
>> small DOS looking program that loads, and come back to a fully installed
>> workstation.
>
> I hit F12 on a server , and it just goes and installs my client as per
> previous instructions. Ok, so I had to tie it to a MAC address, because
> not all machines are the same. I have seen some fancy hacks where someone
> has got PXE to halt, select a template to continue, but that *is not*
> hands free.
no fancy hacks needed for the windows RIS service. It works without hacks,
trial and error or asking people for help.
>> go bother some IT guy to see how it works, and has worked for like 10
>> years now.
>
> Grumpy, arn't you ?
just annoyed at things that are blatantly stupid, like the sun vi bug
where you can't do multiple pastes. Some rock star here traced that one
back to the mid 1980s. It still hasn't been fixed.
Interacting with the sun DHCP server is hokey. ISC's DHCP server isn't
exactly a gem, but it's easy to work with, doesn't use bizarre dead ended
settings and can actually log things in a meaningful way.
then of course, some people like extremely long commands that wrap around
their terminal many times.
>> Oh wait, maybe a team of bored sun people in the UK office wrote a front
>
> From experience, they have no time to be bored, they do stuff in their own
> time
sure they do, like hand holding and explaining bugs that won't ever get
fixed or explaining processes that were never correctly documented, or
writing JET (thanks to those who did) because the jumpstart process is
such a trainwreck and SUN HQ in the USA was too busy renaming every single
product over and over again to pay attention to what they really do.
>> end to the command line for the sun DHCP server that inputs and
> outputs
>> sensible values because the people who wrote the DHCP server never
>> bothered to do it themselves.
>>
>>>> If anybody can explain why a MAC address and hostname have to be used
>>>> at least 50 times per client, I'd love to hear the logic behind that.
>>>
>>> Again, not sure what your problem is here, you supply it *once* to a
>>> JumpStart tool like JET. The tool then handles how and when they get
>>> handed to the client.
>>>
>>> I think 50 is an over exaggeration.
>>
>> ok how about 35 times?
>
> Sorry, bit from my POV, its once.
>
> What happens under the bonnet is not a concern.
what happens under the bonnet is every unix admin's concern when things
are not working.
>> To any nay-sayers, just watch how you net install windows.
>
> I'm still watching, nothing happening
well keep watching and report back when something happens.