Bug 4439

Summary:	annoying and probably unnecessary reboot after partitionning
Product:	Mageia	Reporter:	Morgan Leijström <fri>
Component:	Installer	Assignee:	Pascal Terjan <pterjan>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	Normal	CC:	alien, pterjan, thierry.vignaud
Version:	Cauldron	Keywords:	NEEDINFO, PATCH, Triaged
Target Milestone:	---
Hardware:	i586
OS:	Linux
Whiteboard:
Source RPM:	drakx-installer-stage2	CVE:
Status comment:
Attachments:	do call udevadm in installer

Description Morgan Leijström 2012-02-08 16:13:23 CET

Description of problem:
After making come changes (ex remove a physical partition between other, so it renumbers them then it say it need to reboot to proceed, and use have no other choice.

In some cases it is a big problem, like for me because install using the DVD iso have too many bugs so i have to use boot.iso, and i have slow and expensive internet.  I use urpmi-proxy and squid, but both refuse to cache the <architecture>/install/stage2/*.sqfs   - totall leading to much grief: much waiting, and waste of limited GB/month.


Questions: 
1) does it really need to renumber partitions?
2) does it really need to reboot - it have not begun installing yet...?

Version-Release number of selected component (if applicable):
current today i believe; used mga2 boot.iso and it retrieved it

How reproducible:
Well, i have now rebooted 5 times because of other problems too...

Steps to Reproduce:
1. During install i.e remove a partition between other so it need renumber partitions
2. Press proceed, and it asks to reboot

Comment 1 Manuel Hiebel 2012-02-09 15:14:58 CET

Hi, thanks for reporting this bug.
Assigned to the package maintainer.

(Please set the status to 'assigned' if you are working on it)

Keywords: (none) => Triaged
CC: (none) => pterjan
Assignee: bugsquad => thierry.vignaud
Source RPM: current today i believe; used mga2 boot.iso and it retrieved it => drakx-installer-stage2

Dan Joita 2012-03-07 11:21:14 CET

CC: (none) => djmarian4u
Summary: Annoying and probably unnecessary reboot after partitionning => annoying and probably unnecessary reboot after partitionning

Comment 2 Pascal Terjan 2012-03-07 11:39:15 CET

Partitions don't have a number, their number is their position.
For primary partitions, there are 4 fixed slots (partitions 1 to 4) and number don't change if you remove one of them.
For extended partition (>= 5), this is a list with each partition pointing to the next one. If you remove one in the middle, the position changes.
It should however not ask to reboot unless the kernel refused to reload the partition table.

Dan Joita 2012-03-07 11:46:56 CET

CC: djmarian4u => (none)

Comment 3 Morgan Leijström 2012-03-07 21:51:06 CET

Thank you for the explanation.

Is there a way to tell the cause why it want to reboot next time we see this happen?

Comment 4 Thierry Vignaud 2012-03-15 18:01:15 CET

Maybe because one was mounted

Comment 5 Morgan Leijström 2012-03-15 23:08:04 CET

This was partitionning during install.
I do not think i selected to mount anything.
Does it mount by itself?

Thierry Vignaud 2012-04-18 17:08:35 CEST

Assignee: thierry.vignaud => pterjan

Comment 6 Thierry Vignaud 2012-04-27 12:59:32 CEST

I've seen it once with LVM and/or RAID.
We would need logs from someone who can reproduce it.
Just insert a floppy and type "bug" in tty2.

CC: (none) => thierry.vignaud

Comment 7 Thierry Vignaud 2012-04-27 13:01:20 CEST

Created attachment 2119 [details]
do call udevadm in installer

Just a wild guess, but we still had two missing calls for udevadm (were done for diskdrake in standalone mode, but not in the installer).
However now that we use udev in the installer too, we need to call it.
I'll commit this tonight.

Still having logs would help

Thierry Vignaud 2012-04-27 13:04:08 CEST

Keywords: (none) => NEEDINFO, PATCH

Comment 8 Thierry Vignaud 2012-04-29 14:14:47 CEST

I've commited the fix and a fixed installer should soon be uploaded.
As for urpmi-proxy vs *.sqfs, please open a new bug report against the urpmi-proxy  package.

Status: NEW => RESOLVED
Resolution: (none) => FIXED

Comment 9 AL13N 2012-04-29 23:57:34 CEST

i'm just noting that the sqfs files change often for cauldrons, and thus the cache doesn't help since there's newer ones...

if you would pull out the internet from the machine running the urpmi-proxy, it would use the cached version (HIT_AFTER_FAIL), in case of doubt, go check directly on your urpmi-proxy directory and see if the .sqfs is there.

CC: (none) => alien

Comment 10 Morgan Leijström 2012-04-30 10:45:16 CEST

Thank you Thierry.

OK i found the sqfs files in cache.