I thought Solaris sets up everthing "within" the primary partition and so its swap has got nothing to do with the one outside. I must admit in my current setup Solaris has no access to my other disks as I run my Solaris without my hda and hdc inserted but my two Sata were left in place.

Unless one sets it out to use Linux's swap I couldn't see How Solaris makes the connection with the swap outside its primary partition. In setting itself up these are what the Solaris did with the primary partition

c0d0s0 /
c0d0s1 swap
c0d0s7 /export/home

I have a lot of faith in Grub because it boots every system in my box. It is the simplest and the easiest of all boot loaders. It is also only boot loader that can boot a system manually and therefore allows a user to correct errors in advancing one step at a time.

Solaris by default uses one of its slices as swap space, but nothing forbid to set it to use/add another primary partition for that.

I don't think it can use a logical partiton though.

Your commant was about drak site ,,,,and the link is about how to install more than two OSs on two HDs.





Now my query was:
This query was not related to your comment,,,,it was my question to the forum,,,,,thanks for your answer.

My "dark side" comment was only about several version of Windows 2K on the same machine.

Beware that this link is becaming slightly obsolete with
recent Solaris releases.

I was thinking like you,,,thanks that you confirmed this (obsolete) to me.

By the way,,,Are you aware of any updated link related to same topic?

Does that mean you do not recommend using an extended partition when installing Linux and Solrais together ?

You can use extended partitions along with Solaris, overlapping issues are unrelated to extended partitions presence.

I can't point a finger at Solaris because I had FreeBSD too together with the Linux at the time my partition table got into trouble.

I am convinced the subdivisions of a BSD system are similar to the logical partition in an extended partitions and that causes definite confusion to Grub/Linux for the following reasons

(a) Grub normally boots a partition, say 3rd partition in first disk, by the convention root (hd0,2). BSD can be booted by Grub using root (hd0,a) because the "a" slice is the root of a BSD. I have installed two different versions of BSD in the same disk and they both have a "a" slice. Grub always boots the first one and disregards the second possibility of (hd0,a). Therefore two BSD slices names are not unique to Grub.

(b) The slices inside a BSD system are mounted per /etc/fstab exactly the way Linux mounts other logical partitions using exactly the same filename.

(c) There is sufficient eveidence in the official BSD web site documents to support that there is a conflict between BSD with Linux in the extended partition. Linux also calls the BSD's internal slices as hda9, hda10 etc as the contuation of the existing logical partitions.

I am too green to know how Solaris would react in the similar situation.

If I install the Solaris OS on one primary partition,,what will the advantage of extended partition be ?


Second query,,,Which one should I install first Linux or Solaris ?

An extended partitin can hold about 60 logical partitions! All Linux can be installed and booted from logical partitions.

Depending how you use it you can get over 200 partition out of one hard disk.

FreeBSD and Solaris are using an almost identical way of subdivising a primary partition. The main difference is terminology, Solaris use "Fdisk Partition" for the primary, and either "Slice or Partition" for the inner area, while BSD use "Slice" for the primary partition and "Partition" for the inner one.

Both BSD and Solaris agree to the fact only one primary partition can be accessed, the first one found. Should you have managed to create more, you'll have to change the partition IDs to make the second visible (and loose access to the first). Grub can do that.

Linux has a weak partition standard, and names the partitions sequentially in the order it found them, Solaris, Grub and BSD are using a descriptive naming that give more details about the partition and doesn't break when new partitions are created.

All of this is, in my opinion, unrelated to the error messages that sometimes show up about "partitions not ending at cylinder boundary", "extending past the end of the disk" or "overlapping", which are caused either by partitioner bugs, or O/S disk driver incompatibilities.

The acid test comes when a user has to access data across different platforms by mounting partitions outside the system he/she is running in a PC.

Linux’s logical partitions hda5, hda6….and so on are based on the assumption that there is only one extended partition which can be either from hda1, hda2, had 3 or hda4 depending on the time the first logical partition is formed.

Linux does not have a mechanism to identify uniquely the inner subdivisions of a BSD or Solaris system, unless there is no extended partition in the same disk then Linux can treat these inner subdivisions the same way it treats the logical partitions.

I believe the conflict between the inner “slices” of a BSD/Solaris and the Linux logical partitions to be more than just a bug in the software. At least that isn’t the way the message I got from the BSD web sites.

I don't think it's a bug, just poor design.
Linux can access both extended partitions and BSD/Solaris slices, the issue is that there is no predictible way to identify them from their names.

This is what I did:

I devided HD1 38 GB (c0d0) to three (3) primary partitions, I used partition commander utility V9:

1- 15 GB type: NTFS Win2000 Server

2- 5 GB type: Fat32 to be accessed later by Linux and Solaris.

3- 18 GB type: NTFS 2003 Server.

I installed both 2000 server and 2003 server.


I devided HD2 38 GB (c0d1) to four (4) primary partitions, I used partition commander utility V9:

1- 15 GB type: Solaris (available through partition commander utility)

2- 5 GB type: Fat32 Storage.

3- 500 MB type: Linux swap.

4- 17.5 GB type: Linux Ext3.

I installed Solaris before Linux

During installtion I have these screens:


A-First screen
Preserve data
Do you want to preserve existing data ? At least on of the disk you have selected for installing Solaris software has file systems or unnamed slices that you may want to save.

I pressed "F4_Preserve" button and I received the below screen:


B-Second screen
Preserve data Screen
On this screen you can preserve the data on some or all disk slices. Any slice you preserve will not be touched when Solaris software is installed. If you preserve dat on / (root), /usr, or /var you must rename them because new version of these file systems are created when Solaris software is installed.

WARNING: Preserving an ‘overlap’ slice will not preserve data within it.
To preserve this data, you must explicitly set the mount point name.

MountPoint---------------------Disk/Slice--------------Size
[X]--------------------------------c0d1s2----------------500


Be aware the 500 MB in the line above is same as my Linux swap that I have created before in step 4,,,What does that mean ?

What should I do with this ?



Then I pressed “F2_Continue” instead of “F4_Preserve”.

Then "F2_Auto Layout" button.

Then "F2_Continue" button.


Then I received the screen below:


File System and Disk Layout
Current Disk Partition Information

Part#-----Type------Status-----------Type------------Length
1-------------------EXT_WIN--------30732345----------10233405
2-------------------LINUX Nat------40965750----------1028160
3-------------------LINUX Nat------41993910----------36162315
4---------Active----SOLARIS2-------63----------------30732282
Please select the partition you wish to boot:

Does the LINUX Nat in the screen above refer to Linux swap partition and Linux Ext3 partition that I have created before in step 3 and 4 on HD2 ?

After finishing installation of Solaris (Still I have not installed Linux), I tried to check my installtion, I booted Solaris, I was receiving this message:

Dec 16 03:49:18 svc.startd[7]:svc:/platform/i86pc/eeprom:default:Method or service exit timed out. Killing contract 40.


Dec 16 03:49:18 svc.startd[7]:svc:/platform/i86pc/eeprom:default:Method or service exit timed out. Killing contract 42.

Dec 16 03:49:19 svc.startd[7]:svc:/platform/i86pc/eeprom:default:Method "/usr/sbin/eeprom -I" failed due to signal KILL.
[platform/i86pc/eeprom:default failed (see 'svcs -x' for details)]

What should I do to fix this ?

What is 'svcs -x' ?