Puppy Linux Discussion Forum

[mikeslr]

Hi All,

The problems discussed below may have nothing to do with Puppies; or Puppies may only be indirectly involved. I write out of curiosity, concern and perhaps an over-abundance of caution. Although I have, in part, been able to recover from the problems, it would be assuring to know that Puppies were not involved before I send a Puppy to someone else, who may not be as fortunate as I.
My main computer, including its two hard drives, is less than two years old. Both drives are Western Digital Sata. The one identified by Puppies as sda is 200 Gb. The other, sdb, is 250 Gb. Formerly, as I explored various Linux distributions, I had used gparted to create 6 primary and logical partitions on sda, and 9 primary and logical partitions on sdb. Four of the partitions on sdb were formatted ntfs. The sdb1 partition, a primary, holds the Windows XP OS and its programs, identified when I boot into it, as C:. The three other ntfs partitions hold data and storage relating to XP. The rest of the sdb partitions, all logical, were formated Ext3, intended for exploration of ArchLinux –something I never got around to-- and never used.
Sdb2, a primary formatted as Ext4, holds my various Frugal Puppies, SFSes, Program Folders, and some data folders I frequently access. Sda1 and sda3, also Ext4. also primaries, hold data and sfses infrequently employed, and pets I may want to install without having to download again, as I explore various Puppies. The other 3 partitions previously were used by Pinguy Linux, an Ubuntu derivative.
About two weeks ago I booted into XP and discovered that it no longer recognized the other 3 ntfs partitions. Booting back into “a” Puppy, gparted indicated that both sda and sdb were “unallocated.” Despite that designation, various Puppies were able to access files on any partition.
Fortunately, I was able to boot back into XP, install EaseUs Partition Recovery software, http://www.easeus.com/partition-recovery/, and use it to rebuild the partition tree relating to my primary partitions. EaseUs can recognize a partition formatted as Ext4 as “existing,” but doesn't know what they are. Regarding my Linux partitions, EaseUs reported only the logical partitions as “lost.” Hoping that the problem only related to extended partitions, which I wasn't using anyway, I did not bother to recover them.
Despite that limitation, loss of my computer's partition tables has recurred frequently. It has not occurred on any of my other computers which, I note, use IDE drives. But with the exception of a Thinkpad T42, such other computers are rarely used.
As mentioned above, the problem may not involve Puppies, or involve Puppies only indirectly. I explore Puppies still in the alpha and beta stages of development. I have on more than one occasion resorted to “pulling the plug” out of frustration or impatience when one of those early stage Puppies “hung.” Such hard-shutdowns are not recommended for assuring a hard drive's longevity.
Although I rarely boot into XP, fairly frequently while searching the web I run across some program, file or (using firefox's Save-as-PDF addon) create a file which may pertain to the work or programs I have under XP. When that happens, I download it in Puppy, mount an ntfs partition on sdb, and copy/move it there.
Completing the picture, I've noticed the following: I usually set up Puppy to display drive icons on the desktop. Puppies have to read the various partitions when it creates such access. After EaseUS recreated the partition tables (without the formerly existing logical partitions) Puppies still reproduced icons for the now non-existent (or non-formated) partitions. Running desktop-icon-manager to first discontinue separate icons, and then show them again eliminated the “phantom” icons from the desktop. However, if rox was opened and browsed to /mnt, such “phantoms” could still be seen.
I can not determine which, if any, Puppies are implicated in the loss of partition table problem. Although I discovered the problem two weeks ago, I have explored during the last several months, through their various iterations, Guydog/dpup/Exprimo, Slacko, Lupu/Lucid, precise, and Saluki. The iterations of some of these have involved kernel changes, and –I would guess-- ntfs access utilities changes. Puppies I currently use or explore include Saluki 021 (although 17, 18 or perhaps 19 was in use when the problem appeared), the most recent Slacko (although its second most recent version was in use when the problem surfaced), pemasu's precise, LupuPlus (not present when the problem first appeared), two other versions of Lupu 5.28, Puppy-Lupu-014, Three-Headed-Dog, Exprimo-single-core and two versions Exprimo-Multi-core. Although at the time the problem manifested, I most frequently used Three-Head-dog or Saluki, that might just be a co-incidence as I probably booted into the Puppies at least every other day.
That both “relatively new” sata drives are manifesting problems at the same time is not likely to be a coincidence, or manufacturing problem. If, however, the problem is one of hardware, or my abuse of it, they will have to be replaced. On the other hand, if the problem is related to Puppy, or a Puppy, it is best that we identify and rectify it.
So, is anyone else experiencing partition table problems, and if so, using which Puppy/Puppies, and under what circumstances?

mikesLr

[bigpup]

First where did you get Gparted?
The only one I would use for partitioning like you did is the live CD of Gparted.

Gparted live CD

For partitioning I would suggest you use the Gparted live CD that you can get from here.
You can download a free version to make your own Gparted live CD.
It is up to date and specifically made to run Gparted.
Info:
http://gparted.sourceforge.net/livecd.php
Download:
http://sourceforge.net/projects/gparted/files/gparted-live-stable/

There is more hardware involved than just the hard drives.
Motherboard interface and control chips.
Connection cables.
Power supply.
CPU.

As you say, you have used beta versions of Puppies.
Anything is possible.

[jim3630]

Mikeslr

think you have more puppies than me although we share usage of some. however, i only have linux distros on hdd and only linux files now on the three external harddrives.

have not experienced partition difficulties as described on that scale but have had partitions filling up mysteriously. recently found the culprit in Saluki usage of Thurnar (sic). the filemanager creates a .trash file and raises hell with your savefile space in frugal installs which i have. there is a thread about it somewhere here of late but that could only be a small account if any toward your problem.

hope works out.

[mikeslr]

Thanks bigpup and jim3630 for responding.

Jim, I agree with you. It's probably not Saluki's Thunar. Although Saluki defaults to using Thunar, I installed Rox-Filer and almost always use it. But one of us might suggest to jemimah that, if possible/not too complicated --or already implemented-- the default chooser include the option to set Rox-Filer as the default file-manager.

bigpup, I don't think it's a gparted problem. When I set up the computer 2 years ago, +/-, I used gparted on a Puppy CD to delete Vista from the hard-drive it came with as (if I recall correctly) XP wouldn't otherwise install. I think at that time I also created four primary partitions. Installing XP reformatted the first primary partition. When I added the second drive, I again used a Puppy CD to create its partitions. But that drive is exclusively Linux, with Linux partitions. The last time I made any changes --again using Puppy's gparted-- was about a year ago. Because neither drive presented a problem for 1 to 2 years, I don't think my use of Puppy's gparted is a likely cause.

mikeb pm'd me with several of ideas. Two of them I have already employed by chance. One was to avoid the use of extended partitions which I now understand --my words, not his-- was a "work-around" never really well implemented. The second was to avoid --again, my words-- having Linux mess with ntfs partitions by using of a Fat32 partition for transfers. Ntfs is not as forgiving as Linux partitions. Like you, bigpup, mikeb suggested that the problem might not be as simple as faulty hard-drives. Again like you, one of his suggestions was to check the cable connections. That's something I can do. There were a couple of others which I'll have to re-read, think about, and figure out whether I have the mechanical skills to try.

Again, thanks for responding.

In science, there's no such thing as a "failed experiment." Every experiment tells you something: if nothing else, that you're hypothesis was faulty.

As I suggested in the initial post, the problems may relate exclusively to my system, or how I abused it. My primary purpose for starting this thread was to make certain that those problems were not an epidemic being suffered in silence. I'll wait a few more days to see what develops. But so far, that does not appear to be the case.

mikesLr

[LateAdopter]

A few thoughts on partitions...

1 tazoc had a bug in beta3, I think, that affected ntfs partitions, that didn't mount/unmount properly, If you don't specify psubdir in menu.lst then puppy mounts and unmounts all of the partitions during bootup in order to search for the operating system. This problem may exist in other puppies.

2 If you partition the disk with gparted the partitions are entered in the MBR in the order you create them, not in the order they are physically on the disk. It is valid, but not what other tools do. If you modify anything using Windows disk manager, it reorders the entries in the MBR and so the sda1 etc. identities of the partitions change.

gparted also partitions the disk right to the end, if you let it. It does not, by default, align partitions with cylinders, These are not consistent with the IBM conventions e.g the first and last tracks are reserved.

3 If you install a linux bootloader in a way that modifies thew MBR it may conflict with other tools.

[greengeek]

mikeslr: Jemimah made an interesting comment about partitions being eaten here:
http://www.murga-linux.com/puppy/viewtopic.php?t=73687&start=3013

[greengeek]