viking777

The KDE devs need to realise that the Linux 'mascot' is a penguin, not a lemming. Even if the majority of them want to jump off of this KDE4 cliff, there must be somebody in there that can see how ill advised it is. We need somebody to ditch this unworkable nonsense and start work on KDE 3.6 asap.

FORK OFF!!

That is what I say :lol2:

Hi, viking777,

at least when using fdisk the first logical partition within an extended partition is always /dev/sda5

http://tldp.org/HOWTO/Partition/fdisk_partitioning.html

Checkout example 5.4

...

Thanks for that scarecrow, I was aware of the 4 primary limitation, but I was not aware that meant that all logicals started with #5 even if you don't have the four primaries. Of course in my case it was just coincidence that my first logical partition happened to be #5 anyway.

fopetesl - aplogies for interfering in your thread. I hope you are able to sort out your problem with the information that scarecrow has given you.

Since /dev/sda3 is an extended partition containing logical volumes, the first logical one would be /dev/sda5

I don't want to hijack this guys thread scarecrow, but your statement there puzzles me. If /dev/sda3 is your extended partition then surely /dev/sda4 would be the first logical partition on it (which it is). Why should it be /dev/sda5? I'm not saying you are wrong or anything, I don't really know, just asking.

On my system /dev/sda4 is the extended and /dev/sda5 is the first logical partition which seems - well, logical - if you will excuse the pun!

Now Mandriva won't boot, comes up with a fault on /dev/sda5 (the new NTFS partition), then throws me into a login prompt which will not accept anything.

I think we are going to need some more information about this statement, partitcularly the login prompt that 'will not accept anything'. If you really have a login prompt then it should accept your username and password, if that isn't happening then what is? When you get to this unhelpful login prompt have you tried pressing CTL/ALT/F1 (or F2 etc) to open another login window?

What information do you have on the fault on /dev/sda5? Anything? Nothing?

From my own experience I have found that every time I mess around with partitions it always helps to reinstall grub afterwards. This might sound difficult, but it really isn't. There is nobody in the world more command line phobic than me and I can do it easily.

Of course that brings me back to my original question. If your login prompt really will not accept anything then you will not be able to easily alter grub, but we really need a bit more information.

I just had a look at those screen shots you posted and I noticed 1 consistent error which is a read error from sr0. Earlier in the screen shot you will see that sr0 is how your cd/dvd is mounted. If it is getting read errors from your drive then that could possibly mean a faulty disc. I am not sure about this because I don't know enough about it, but it is one possible source of errors.

If this is the case then you might look at reburning the CD you are using, possibly at a lower speed, to see if that makes a difference.

I stress though that this is only guesswork, I in no way claim that a reburn will definitely solve your problem, it is just a strategy you might try if there are no other suggestions.

The plot thickens!!

Why is his / partition on /dev/sdb1? That surely means it is on a second hard disk? Does the machine have two disks? Has it been installed to some external device?

Or is it that - as usual - I don't know what I am talking about??

If you have a live cd that you are installing from then boot into that, start Mandriva Control Centre, select 'Boot' followed by 'Setup boot system' and follow through the wizard to reinstall grub.

Alternatives are 'Super Grub Disk' if you have a copy of that, or another install - and probably a lot more I haven't thought of or don't know about!

Disk is 400GB dedicated to Mandriva Spring so not sure what the problem is

The thing is if The Berr has a 400gb disk with nothing on it but Mandriva then it is difficult to understand how a 4.4gb /var file is going to stop it from booting. It surely has to be something else?

Your last screen pic may have a little more to do with the actual problem although I am afraid I can't diagnose it, perhaps Greg or someone else can help you with that.

EDIT. Correction to that last statement, I just ran 'du' on my own system proc file and it came up with similar errors so that probably has nothing to do with it either.

I can only think that you need to repair the file system with e2fsck. (While I am thinking about that, on my last post I suggested you ran the command 'mount' to ascertain your root partition number. Of course this would only work if your partitions are actually mounted at the time and I am not sure if that is the case. An alternative command that will reveal similar information is 'fdisk -l')

Unless of course you are using a laptop and want suspend/hibernate, then you'll definitely need double the ram for it to work - especially suspend to ram! ;)

That is interesting, because I have 2gb ram and 2gb swap on my new laptop and suspend works perfectly for me - not that I use it much, but I have tested it and it does work.

Ok, when you get to the screen that you posted, do what is says and give the root password. having done that type the command

mount

this will give you the device location of your / partition (if you already know it then ignore this step), you should get something like this:

/dev/sda6 on / type ext3 (rw)

none on /proc type proc (rw)

/dev/sda5 on /media/mandriva2009 type ext3 (rw)

You can see from my example that my / partition is on /dev/sda6.

Armed with this information you can then run the command

dumpe2fs /dev/sda6|grep -i superblock

Of course you will substitute 'dev/sda6' for the location of your own / partition. (NB the | symbol between dev/sda6 and grep is accessed by pressing shift+\ on a UK keyboard - different if you are on another keyboard layout)

This should give you a return something ike this

Primary superblock at 0, Group descriptors at 1-2

Backup superblock at 32768, Group descriptors at 32769-32770

Backup superblock at 98304, Group descriptors at 98305-98306

Backup superblock at 163840, Group descriptors at 163841-163842

Backup superblock at 229376, Group descriptors at 229377-229378

Backup superblock at 294912, Group descriptors at 294913-294914

Backup superblock at 819200, Group descriptors at 819201-819202

Backup superblock at 884736, Group descriptors at 884737-884738

Backup superblock at 1605632, Group descriptors at 1605633-1605634

Backup superblock at 2654208, Group descriptors at 2654209-2654210

Backup superblock at 4096000, Group descriptors at 4096001-4096002

You can now do what the error message is telling you to which is to run

e2fsck -b 32768 /dev/sda6

Again substituting any numbers that are relevant to your installation. (go right through the list if you need to)

As you can see from my example Linux saves several backups of its 'superblock' and any one of these will be enough to check and repair the disk - if , and it is a big if, you are actually dealing with the right partition in the first place, because as somebody else said, if your /etc/fstab is messed up and you are trying to access the wrong partition then none of this is going to work.

Anyway, give it a try and let us know what happens.

OK Let's make the rather foolish assumption that the error message that you are getting actually has something to do with the real error. It may not have, but we have to start somewhere.

When you get to your console, login as root and type:

drakconf

Use a combination of the tab keys and the up/down arrow keys to highlight 'bootloader'.

Press 'Enter'

Use the tab key to move to 'Advanced'.

Use the up/down arrow keys to highlight 'clean /tmp at each boot' (I have no idea why this is not done by default)

Press the space key to select it (an x should appear in the brackets)

Use the tab key to select 'close'

Press enter.

Use the arrow keys to select 'next'

Press enter

Use arrow keys to highlight 'finish'

Press Enter.

type

shutdown -h now

Then restart and if a full /tmp file was really the problem then you should be able to continue normally. Of course if it wasn't????

This problem is now solved - I suppose I should tell you how I did it - except I don't really know any more than I know how I got into the mess in the first place.

I guess the key was to use the 'testdisk' program again, although it is extremely nerve wracking as it has very little information with it and I am not at all sure what it is doing. Anyway I used it to create a new partition layout (which was roughly the way it was before this all started except that my Mandriva partition - the one that I restored - was gone). I wrote this to the mbr.

Unsurprisingly after that nothing booted so I then used Puppy linux to manually edit all the entries in menu.lst. This resulted in two of the 4 operating systems booting again. I then used Acronis to reinstall the missing Mandriva partition (strangely enough Acronis could see where the missing partition fitted and just popped it straight back in there). Unfortunately in doing so it renumbered all the partitions again so I was back into Puppy to once again manually alter the menu.lst entries.

Again nothing booted so it was into Super Grub disc to try to reinstall grub. Both the automatic and the manual methods it offered to do this failed so in the end I just got into a grub prompt and did it by hand and this worked first time.

This then left me with Vista unbootable, but this was cured by running the Vista recovery disc on automatic repair twice!

The only thing left unworkable then was my swap file for some reason, so back into system recovery to run mkswap and this was fixed as well.

Everything is now working 100% and I have full access with all my disk management tools once again.

I think I will switch the computer off now before I do any more damage.

Thanks for that reply David, but I certainly don't want to go down that road just yet. The hard drive, and the computer itself, is only 3 weeks old, it really would be unfortunate if the hard drive were failing already, this is definitely something that I have done to it by using windows tools to restore Linux partitions. It has always worked before, but recently it has become a real problem.

I am not absolutely sure how it happened, I was just installing a backup image of a partition that I had messed up when all hell broke loose! After the reinstallation all I got was grub errors. I used all the tools that I had to recover from these and I have been successful. I have 4 OS's on my current disk and all of them boot normally. Trouble is though that all my disk management tools, command line or gui just report errors in the partition table (or just don't work at all).

For example fdisk just says " Unable to seek on /dev/sda", gparted just reports that the entire disk is 'unallocated', diskdrake just says 'partition table is too corrupted for me to read' (although it does go on to show the disk contents but warns that I will not be able to make any changes to any of the partitions). The only tool that does not report any errors is the windows tool Acronis disk director.

The problem is though that Acronis and the Linux tools both report different disk arrangements. Eg. Acronis says my Mandriva 2008.1 partition is /dev/sda5, whereas diskdrake says it is /dev/sda9. The only partitions that haven't changed number are the two windows partitions and the Cooker partition (although according to diskdrake it has mysteriously moved itself from the middle of the disk to the end!). I am amazed I got it to boot at all given the mess it appears to be in.

I have tried SysRescue cd, parted magic cd, super grub disk, knoppix, Acronis, testdisk and both a Mandriva and a Suse install disk to try and remedy this situation but I haven't succeeded.

Obviously the partition table is corrupted, but how can I fix that without deleting everything and starting over? I don't really want to do anything destructive particularly as everything boots, but not being able to use a single disk management tool is a limitation that I couldn't live with for long.

Oh, the good news is that the partition I restored is working perfectly btw, it just is no longer in the place where it used to be.

This is just a thought, but have you tried paying it with Kaffeine?

Or you could always install VLC media player and try it with that. VLC plays just about anything.

Music to my ears my friend!

During the course of my life I have kind of got used to usually being in a minority of one when it comes to my ideas, but to think there is actually somebody out there that thinks the same way is refreshing!!

In Package manager search for 'codecs' you probably need the W32 package. If that doesn't work try any others you find there that you think might work. (You may have guessed I am not an expert!)

What Ian says.

But if you can install from tarballs you are not so much of a noob as you make out!!

Hope you enjoy Mandy anyway, it is a good choice.

We just have to wait for Acronis to sort themselves out. Did you email them about it?

I did. As I said in my previous post they emailed me a link to their $700 Linux server backup software, and then asked me if their solution had solved my problem and would I like to take part in a customer feedback survey. :unsure:

I've no experience of the Ferrari (computer or motor!!) but my Aspire 5920 is pretty near as good as you could get when it comes to Linux compatibility.

I found an update to FF3 last night which has improved the font rendering a lot, but they still look better in Opera, maybe I have found a new favourite browser?

tar gives you a reasonable backup I am sure (although I don't use it) but what I want, and used to have, until inode sizes were messed around with, is a restorable, resizeable, image in the least possible time and the least possible space with the capability of incremental additions and all with just a couple of mouse clicks. Afaik this is not yet possible with Linux but it used to be with Acronis.

I guess I will just have to put up with its sector by sector copying mode until something sensible comes along in the Linux world - if it ever does.

I don't have to post any links nexus, you are looking at one if you are reading this forum, it looks awful, with scratchy broken characters and out of focus icons with jagged edges - truly horrible. It is only FF3 that is looking like this btw all other programs are normal as is playing a dvd with kaffeine and if I open web pages in konqueror they are perfectly OK as well. I have tried all the font settings in the ff 'preferences' menu with about half a dozen different fonts at half a dozen different sizes and this is all I get. I have also installed msttf fonts and they make no difference either.

If you google for 'firefox 3 fonts' you will see that I am not the only one with the problem.

Are you sure you really wan't to fisher? I am using it now and it really sucks. The font rendering has to be seen to be believed - dreadful.

I am presuming that since these changes are being adopted by various distros - as there must be a reason for changing the inode size, and thus, Acronis will most likely update this for their newer versions. At least, something I would expect of a commercial product.

 

You shouldn't be forced at least to create your partitions exactly how Acronis want :)

Your expectations are absolutely correct ian, but the reality is a little different. In order to try and beat this problem I upgraded from True Image 9 to version 11. It has made no difference. It still images any Linux drive that I have, but it only does so on a sector by sector basis which means that it takes longer, gives a larger image size, and does not have the ability to resize a restore. As an experiment I created my latest 2008.1 install with a partition created by Acronis disk director and this partition it will image normally.

The point is though that I should not have to - and don't want to - use commercial disk imaging software, a satisfactory solution should exist for Linux, but it doesn't, at least not unless someone has produced something in the last 3 weeks which is how long ago I last tested the alternatives. (The last one I tested was Clonezilla - I did an image/restore on my old laptop. It trashed the drive I saved the image to, it trashed the drive I restored the image to and it trashed the bootloader as well - pretty effective eh? Now the point is that I have the knowledge and the tools to put right all these disasters, but an awful lot of people won't have.)

Backups and imaging are so so important to every computer user that a sensible imaging solution should be one of the first packages ever produced for any Linux distro but it never is, we always have to suffer the likes of dd and partimage or use commercial alternatives like Acronis which now no longer work.

As I said I would I contacted Acronis support to describe this problem and I even provided links to the e2fsprogs pages describing the changes to the filesystem that Greg2 gave me. As I expected the reply I received was a link to their 700 dollar/pound server product for Linux.

I fully realise that I have no right to criticise anyone since I have never written a single line of computer code in my life and almost certainly never will, I was just born a bit too early to have the chance to do that. But at the risk of sounding like the proverbial worn out gramophone record, I will repeat that all the vast resources that have been thrown at Beryl/Compiz and now the same resources that are being thrown at the Vista clone - KDE4, would be vastly better spent on providing the absolute basics that are required for any operating system and a satisfactory imaging solution most definitely comes into that category.

Please forgive my cynicism.