Bug 7155 - partition order in kickstart
Summary: partition order in kickstart
Alias: None
Product: Red Hat Linux
Classification: Retired
Component: installer
Version: 6.1
Hardware: i386
OS: Linux
Target Milestone: ---
Assignee: Matt Wilson
QA Contact:
Depends On:
TreeView+ depends on / blocked
Reported: 1999-11-19 21:13 UTC by Vilius Puidokas
Modified: 2008-05-01 15:37 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Last Closed: 2000-02-22 17:16:05 UTC

Attachments (Terms of Use)

Description Vilius Puidokas 1999-11-19 21:13:53 UTC
i have in ks.cfg:

clearpart --all
part / --size 128
part /var --size 128
part swap --size 125
part /swapfil --size 125
part /usr --size 2000
part /tmp --size 128
part /export --size 100 --grow

what was most unexpected to get:

/dev/hda6 on /
/dev/hda5 on /var
/dev/hda1 on /usr
/dev/hda7 on /tmp
/dev/hda10 on /export

as we see, the partition order listed in ks.cfg and what i get after
install is not the same. why is that?

problems we get:
LILO fails, because / is on > 1024 cyl.
i have to guess which partition is / when trying to do rescue.

(btw, what was the argument not to include simple user-friendly rescue.img
to /images?).


Comment 1 Jay Turner 1999-11-22 15:09:59 UTC
This issue has been forwarded to a developer for further action.

As for rescue mode, we included that on the installation media, so instead of
having to have a system-specific boot disk and the rescue.img disk, you just
have to type "linux rescue" at the boot prompt.  This will load the rescue image
from the installation source and allow you to use the rescue functions.

Comment 2 Vilius Puidokas 1999-11-22 16:31:59 UTC
(a little workaround i've found)
I noticed that bug is about maxof(--size). looks like hda1 is always largest
partition (non --grow).
say i do:
part / --size 250  ; i'm making it largest
 part /var --size 100
 part swap --size 100
 part /swapfil --size 100
part /usr --size 200 --grow --maxsize 800 ; (--maxsize is buggy also)
 part /tmp --size 100
 part /export --size 100 --grow

now, hda1 is /

Comment 3 Ron Ginger 1999-11-24 20:16:59 UTC
if you make a partition /boot kickstart will put it in /dev/hda1 and lilo can
boot ok.

It would be better if kickstart simply made partitions in the order specified in
the file. Why do you sort it to make the largest first?

Comment 4 Christian Hechelmann 1999-11-26 12:15:59 UTC
For now you can use this patch. This behaviour has annoyed me since DiskDruid/
kickstart has been introduced. When kickstarting machines, I really have an
idea on *why* putting partitions in a certain order.

If at least this behaviour would be manually configurable for kickstart
installs... Something along the lines of

%option libfdisk -sizesort

or similar.

Maybe adding some flags attribute to struct fdisk_partition would be a good
idea, so that one can expicitly specify if the partition should be sorted by
size or not. We could use this flags attribute for other things too.

--- anaconda/libfdisk/partspec.c.nosizesort	Fri Nov 26 13:03:44 1999
+++ anaconda/libfdisk/partspec.c	Fri Nov 26 13:07:40 1999
@@ -101,10 +101,16 @@
 		f = 1;
 	    else if (pri1 > pri2)
 		f = -1;
+/* Sorting by size is *really* annoying and a PITA when doing kickstart
+   since I order the partitions for a reason, e.g. swap areas at the end of a
+   disk are just plain stupid. */
+#if 0
 	    else {
 		fdiskGetConstraint(&p2->size,&csize2,&lsize2,&msize2,&act2 );
 		f = (lsize1 < lsize2);

 	    if (f > 0) {
 		memcpy(&tmpe, &spec->entry[i], sizeof(PartitionSpecEntry));

Comment 5 Jay Turner 1999-11-29 17:17:59 UTC
This issue has been forwarded to a developer for further action.

Comment 6 Jay Turner 2000-02-22 17:15:59 UTC
We have added an "--onpart" argument to kickstart which will enable you to
specify which partition will get each partition that you are specifying in the
ks.cfg file.  This will help to avoid the problem that you are having.  In
addition, the drive space placement argorithm has been improved and now places
the /boot or root (if /boot is not separate partition) below the 1024 cylinder

Note You need to log in before you can comment on or make changes to this bug.