Unusual Jumpstart Question

Unusual Jumpstart Question

Post by Richard B. Joh » Wed, 10 Sep 1997 04:00:00



Howdy.

We have been using Jumpstart as the means for automating the loading of
our servers and workstations (SS20s and SS5s) for well over a year with
great success.  However, there is one situation that has always been
nagging me, and maybe someone out there can shed a little light on the
answer, or point me to where I can find the answer.

We typically reload systems every several weeks since I work in a testbed
and do systems and software integration work.  Thus, the disk partition
and configuration information rarely changes, the packages and patches on
the OS rarely change, the COTS software rarely changes and it's generally
the developer's application software that changes.

So, my questions are:  If Jumpstart is used to completely reload a system
where the OS is not different from the previous load and the disk and
partitioning information has not changed, does Jumpstart perform an actual
repartitioning of the disks and a newfs and fsck, or does the utility
merely check to see if the existing information on the disks is the same
or different from that in the "profile" file?

I have had situations where I used our automated load procedure, and at
the comletion the system came up as the system I had previously loaded.
This has led me to believe that I cannot rely on Jumpstart to "clean" my
system and so now I recommend to the people on my project that they boot
single user off of a CDROM and "munge" the old filesystem by
repartitioning the drives to something different than that of the
Jumpstart installation and then newfs these new partitions.

This adds a lot more time to the installation process, but I haven't had
any more "ghost" systems creep up in the testbed anymore.

Any comments?

Regards,

Rich Johns

 
 
 

Unusual Jumpstart Question

Post by Andrey Ryzho » Thu, 11 Sep 1997 04:00:00



[...]
> So, my questions are:  If Jumpstart is used to completely reload a system
> where the OS is not different from the previous load and the disk and
> partitioning information has not changed, does Jumpstart perform an actual
> repartitioning of the disks and a newfs and fsck, or does the utility
> merely check to see if the existing information on the disks is the same
> or different from that in the "profile" file?

No, it's not that smart. Does everything from scratch.
if you have "install_type initial_install" in the profile,
it *is* an initial install. New label i s written to the disk,
no matter whether it is the same as before.
Try to powercycle the client when it just started creating the
root filesystem, and you'll see that the OS on the disk is
overwritten.

Quote:> I have had situations where I used our automated load procedure, and at
> the comletion the system came up as the system I had previously loaded.
> This has led me to believe that I cannot rely on Jumpstart to "clean" my
> system and so now I recommend to the people on my project that they boot
> single user off of a CDROM and "munge" the old filesystem by
> repartitioning the drives to something different than that of the
> Jumpstart installation and then newfs these new partitions.

> This adds a lot more time to the installation process, but I haven't had
> any more "ghost" systems creep up in the testbed anymore.

> Any comments?

Why not create a file after the Jumpstart is done,
and check if that file exists after another Jumpstart?
Will the disappearance of that file be a good proof?

Regards,
Andrey

 
 
 

1. Unusual Stacksize Question

Ok, this question is kind of strange, and I am not even sure how to phrase
it clearly:

Are there any possible issues with processes that have ELS (Extremely
Large Stackspace [TM])?  [Ok, I made that term up..].  I have an ELS
process that I was tinkering with..  Purely for academic reasons, mind
you, I created huge arrays of doubles and managed to get the thing to
coredump.  I used setrlimit() to set the stacksize limit to infinity.  No
more core dumps.  But guess what?  Like half the time I now get a kernel
panic screen dump and the system immediately hangs...  I should think that
really, as long as you have enough memory, both real and imagined (I made
that term up too), nothing too bad can happen beyond a coredump maybe.

At any rate, I am suspicious of the motherboard and ram this system uses,
but I was just wondering if you guys knew of any issues in the kernel with
extremely large stack segments.. possibly in the context switch code or
somesuch (again, I should think not.. but I figured I should ask
anyway..).

The kernel is the stock redhat 7.1 kernel: 2.4.2-2

I apologize for the randomness of this question.. but at this point I am
trying to eliminate possibilities and if there were a known issue with
huge stack segments in userspace I would appreciate the info...

-Calin

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

2. Telnet into SCO - Term Type Problems

3. Netscape and pine...unusual question

4. Amusements

5. Unusual Kpppd question for Linux Gods!

6. top (procps-0.97) is broken (0.96 OK)

7. Unusual HW question on Creative AWE32 ISA

8. Xfree source code

9. Newbie question : unusual paper format for printer

10. Unusual Routing Question

11. Unusual Mouse Question

12. *unusual* modem-related question ;-)

13. Unusual Question: Anyone tried remote shutdown of a Linux server???