Why is Puppy so unstable?

[silverojo]

I've used various versions of Puppy from 2.0 to 4.1, but I don't dare do a traditional installation of them because they're unbelievably unstable. So I limit myself to the frugal installation, because I can reboot in Windows, delete the pup_save file, and replace it with a working version that I keep on CD for this very frequent problem.

I thought perhaps it's because I use Opera, and Opera/Linux's support for Flash isn't good. So I used Firefox instead--exact same problem, except that flash videos play better.

Whether it's a browser or a game like xshisen, I've never seen how an operating system can completely collapse when one program crashes.

Sometimes I can get by when it prompts me to run the xorgwizard and then run xwin. Usually, though, that's the first sign that Puppy is going to give up the ghost, and at some point soon I'll get the "no space left on device" message that madly repeats ad infinitum, until I press the restart button on my PC, boot into Windows, and replace that pup_save file with the good one from CD.

I've had to do this at least 50 times in the last 6 months. As much as I like Puppy, I simply can't risk relying on it full-time when it's this stunningly unstable and unreliable.

Are there any plans to make Puppy more stable, so that one program crashing won't bring the entire operating system down to its knees?

[Lobster]

You are running out of personal storage
Keep an eye on how much is free (blue box bottom right)
Resize from menu here:
Utility / resize personal storage size

This is they way Puppy works
It is different and not seamless

[WhoDo]

I've had to do this at least 50 times in the last 6 months. As much as I like Puppy, I simply can't risk relying on it full-time when it's this stunningly unstable and unreliable.

Are there any plans to make Puppy more stable, so that one program crashing won't bring the entire operating system down to its knees?

In addition to what Lobster has offered, I'd venture a guess that the problems you are having probably stem from the fact you are storing your pup_save.2fs file in an NTFS partition. That's NOT Puppy's fault; it's not yours either, but Micro$oft could do a whole lot more to stop its filesystem from being a source of corruption. It won't of course.

To ensure your pup_save.2fs file remains corruption free, make sure you regularly run chkdisk /R on your NTFS partition. A pup_save.2fs file is going to be one of the largest single files in any NTFS file system, and so more susceptible than most to a flakey NTFS.

You can almost GUARANTEE the stability of your pup_save.2fs by creating an ext2/ext3 partition for it on your hard drive and saving it there, or saving it on a USB drive that uses either FAT or ext2 format.

I hope that helps.

[Bruce B]

Silverojo,

I'm not one of the cult minded who things Puppy is beyond criticism or improvement. So yeah, if Puppy has a problem I want to square up with it.

It is not supposed to crash.

Through all its iterations, I've not found one unstable one. This experience covers a variety of computers.

The exception I have found are with other people in my support system, who turn it off like a television set. Pretty soon they get an corrupt filesystem and stability issues.

If the problem is lack of swap device or full user file, that is not Puppy, rather user engineering.

Here is current uptime: up 23:06

The reason it is only 23:06 hours is I did a reboot which wasn't necessary, but I wanted to test something.

If I have crashes, lockups I sincerely consider it a problem to fix. And one that can be fixed.

02:21:11 , load average: 0.05, 0.31, 0.25

With my main machine down, I'm using one with 526MB RAM, many will discourage you from using a paging device

Let's see what's going on with that:

Code: Select all

             total       used       free     shared    buffers     cached
Mem:        515964     441400      74564          0      55644     211944
-/+ buffers/cache:     173812     342152
Swap:        51192       4488      46704

Linux wanted to use 4.5 MB for paging at some point in the past 23 hours. It is only a 50mb device, but give Linux one, in the event it wants to page.

Not wanting to 'make you wrong' of course, I believe what you say, but sincerely it is not normal.

With enough specifics, if you can provide them, I'll try and offer any help that comes to mind.

Bruce

[lapis]

NTFS partition. That's NOT Puppy's fault; it's not yours either, but Micro$oft could do a whole lot more to stop its filesystem from being a source of corruption. It won't of course.

Is that really a valid point of view? I may be no expert on Puppy or Linux like you guys but isn't NTFS one of the best file systems rather than the worst like your comment makes out?.

No crashes I have ever had could be blamed on my using NTFS.

That said, I do get situations when browsing where Puppy freezes without warning. This has never happened with any other program and my gut feeling is that it is related to Flash(which I believe is evil - not Microsoft)

Anyway, can I make a suggestion. I apologise if it is obvious. I have found better success with flash and browsing if I point the Firefox cache to /tmp. Even so, Firefox has huge memory leaks when certain types of webpages or flash are used. Watching the Firefox memory usage and closing it when it gets to a certain level seems to be the best insurance against a crash. If I routinely do that then I can have up times of weeks and that is with 256mb of ram, no swap file and about 30mb free on the pup_save. Plus I am still only using the Puppy 2 versions, I presume the newer Puppy's are better?

[Colonel Panic]

Are you using one of the Firefoxes which has been specially compiled for Puppy? I'm using Firepup at the moment and it's pretty good, and fast; I've not had any problems with it so far.

[Indy]

Hi silverojo,

It seems I've struggled with the same problems you've had, especially with the "no space left on device" problem, if not to the same severity.

In fact, just yesterday I thought it was finally time to refresh (or upgrade) my main workhorse 2.15CE setup (which I've been happily running as a LiveCD on my main desktop since the version came out).

But I found this link: HowTo fix "no space left on device" with Opera (a fix proposed by MU).

http://murga-linux.com/puppy/viewtopic.php?t=29707

I followed the advice and did the following:

Solution:
add this line to
/root/.xinitrc
after the line "#!/bin/sh":
Code:
find /initrd/pup_rw/$HOME/.opera -type f -name '.wh.[^dir][a-z]*[A-Z]*[0-9]*' -exec rm -f {} ';'

That cleans up .wh (whiteout) files left by Opera in the .opera directory.

Further down that thread they advise to do something similar for gftp, like this:

find /initrd/pup_rw/$HOME/.gftp -type f -name '.wh.[^dir][a-z]*[A-Z]*[0-9]*' -exec rm -f {} ';'

I tried both BUT, my Puppy was still acting up.

In the end, I ran the following in a terminal (I took out /$HOME/):

Code: Select all

find /initrd/pup_rw -type f -name '.wh.[^dir][a-z]*[A-Z]*[0-9]*' -exec rm -f {} ';'

That took a while to run (maybe 10-15 minutes? I went for a coffee.)

After that my Puppy seemed to start behaving itself. So I put that code in .xinitrc.

WARNING: I don't fully understand the ramifications of taking out "$HOME/" from that command line. All I know is this found a lot more .wh files (thousands???), especially "sqlite" ones. I may have really borked my Puppy, I don't know, but what I do know is I haven't seen the "no space left on device" error since (about 10 hours of use now) whereas before I was at the stage where I couldn't go five minutes before the dreaded message would come up and I'd have to reboot. Maybe someone can tell me whether or not I've got a ticking bomb Puppy now? I do have good backups though so I wasn't too scared to try this. Anyway, have a look at that link. It may be the answer to your instability issues.

[Indy]

Oops! Sorry, correction. I added the code to /etc/rc.d/rc.shutdown (as MU did) so the code cleans up the .wh files at shutdown (or reboot?)

[dejan555]

Is that really a valid point of view? I may be no expert on Puppy or Linux like you guys but isn't NTFS one of the best file systems rather than the worst like your comment makes out?.

Oh, dear god, no!
It's not a subjective opinion if that's what you meant. You've heard about defragmenting fuss, maybe did that on ntfs? Well ext3 is a filesystem that doesn't need defragmenting, and there's less chance to get a corrupted files than when using ntfs

[computerophil]

I love Puppy and I couldn't expirenze that it would be instable.
I do have my favorites and that are 2.12 and also 4.x
Puppy has like everything else pros and contras but i have an old Pentium II with Puppy 2.12. On this computer I don't have an other Operating System.
It is working fine now for 2 years and nowbody would imagine that it is an that old computer.
So maybe it doesn't work fine with you because an other operating system is interfering in some way or the other.

[WhoDo]

NTFS partition. That's NOT Puppy's fault; it's not yours either, but Micro$oft could do a whole lot more to stop its filesystem from being a source of corruption. It won't of course.

Is that really a valid point of view? I may be no expert on Puppy or Linux like you guys but isn't NTFS one of the best file systems rather than the worst like your comment makes out?.

No crashes I have ever had could be blamed on my using NTFS.

Unintended reading. There is nothing inherently wrong with NTFS when used solely for NT-based systems. It has a hotfix capability that overcomes minor file system corruptions without user intervention. That's good IF you are using only NT-based systems but NOT if you are storing another file system within the NTFS.

I have been using NTFS since before it was called NTFS. As any VMS sysop will tell you, NTFS looks so much like VMS in its underlying architecture that it's hard not to jump to conclusions; especially when you remember that the guy who headed the NT development effort was the guy who managed the VMS development for DEC.

My point was that compatibility has never been high, or even necessarily existent, on Micro$oft's list of priorities. Why make it easier for rival OS's to interoperate and take some of your market share away?

NTFS is a particularly fault-tolerant file system, but it really only tolerates its own faults! Puppy uses ntfs-3g to allow Puppy Linux users to access NTFS partitions, but the problems arise when saving data from the ext2/ext3 partition to the NTFS partition AND there is some fault tolerance covering a corruption. You can lose the transferred data AND corrupt your NTFS partition. That is why I don't recommend storing the pup_save.2fs (a squash file system in a single file) inside your NTFS partition.

That said, I do get situations when browsing where Puppy freezes without warning. This has never happened with any other program and my gut feeling is that it is related to Flash(which I believe is evil - not Microsoft)

Yes, since Adobe acquired Macromedia (and Flash), I haven't been impressed with the quality of their output. I was at an Adobe-sponsored educational seminar recently where all of the flagship Adobe products were showcased, and it was just one foul up after another! Not a good advertisement for Adobe IMHO

Anyway, can I make a suggestion. I apologise if it is obvious. I have found better success with flash and browsing if I point the Firefox cache to /tmp. Even so, Firefox has huge memory leaks when certain types of webpages or flash are used. Watching the Firefox memory usage and closing it when it gets to a certain level seems to be the best insurance against a crash. If I routinely do that then I can have up times of weeks and that is with 256mb of ram, no swap file and about 30mb free on the pup_save. Plus I am still only using the Puppy 2 versions, I presume the newer Puppy's are better?

Not a bad suggestion. It's all worth a try. Browser cache is one place where pup_save.2fs files can fill up without the user being all that aware it's happening and then CRASH! It is also important to remember that pup_save.2fs files that have crashed can sometimes be recovered by Puppy, which will mount them with a click in later versions and allow the important content to be copied elsewhere.

Hope that helps.

[Bruce B]

Dear Interested Parties,

NTFS and people who put Puppy on NTFS.

I never suggested or told anyone to do it.

Puppy mounts, to the best of my knowledge NTFS read - write. I'm not sure why. One would likely think if NTFS is the host partition for read-write Puppy files it must be mounted read-write. Personally, I wonder about this.

Do Puppy's engineers 'need' to mount the host NTFS read and write?

There was a time, not long ago, when NTFS was not considered 'safe' to write to, so what did Puppy do? It mounted NTFS read only. It read and wrote to the pup_save file (which was typically called PUP001) on a read only host partition.

If anyone wants to protect their precious Windows, which they paid money for, I don't see why the NTFS puppy host partition can't be mounted read only, the idea I propose is based completely on how Puppy used to do it.

Bruce

Puppy engineers, please comment on this. An inquiring mind wants to know.

[ttuuxxx]

If you want a stable puppy this is how you do it
first format your hard drive to ext3, then do a frual install and reboot and make a pupsave file.
Next install this updated pupsave file, only after you made one by rebooting.
http://www.murga-linux.com/puppy/viewto ... h&id=14644
now go to /Menu/Utility/Resize personal storage file
and now you have my ext3 version to really increase the size, I usually make my pupsave file at 15GB, I've been doing it for a few months now and its never failed me once, make sure its a ext3 partition and not windows, because windows wasn't made for that size of files.
now reboot, it will take a bit of time to make a say 15Gigs pupsave, on my fast pc around 5 mins or a bt less, older computer maybe 10mins.
Now when you get to your desktop you should have about 15+gigs of memory left:) I usually run my pc for about 2 weeks before I fill it up, but most users would use that in a year, lol I'm always compiling stuff.

Next if you don't know how to compile, do a forum search, there is tons of how to compile, for new users.
The thing about compiling is that, when you do so, it builds the software specifically for your system, or Os version your using, Like using some series 3 files on series 4 isn't good sometimes you'll get crashes, Just like when you download Firefox from Mozilla, Thats usually compiled on redhat on a very new pc that isn't related to puppy or your pc what so ever, hence the crash boom bang, oh crap puppy crashed again, well thats not puppies fault, thats Mozilla's. Same with the game your playing where did you get it, ? was it compiled on the same series as your using? Adobe are bastards when it comes to there products, They use bleading edge Glibc to compile flash, if you OS is more than 5 months old forget getting the latest, because it won't be stable. Even big name OS like Slackware really only update Glibc every year or two. So there are only a couple of true puppy compiled browsers, Firepup is one that i did, and also I did the latest Seamonkey, both are really stable and flash works fullscreen on youtube There's also a new FireFox 3 series in the software section section which this other user says he compiled it also, I haven't tried it, but some of his dependencies are a bit strange.
Its always best to compile your software, not just for stability but for the added Linux knowledge which is always good

If you having issues with flash, keep trying different ones, until you get one that works best for you pc
ttuuxxx

[Bruce B]

ttuuxxx,

I wish to mention that Puppy doesn't make my pup_save files, I do and I have for a long time.

dd to make the size you want
then format

Puppy will pick it up by the name and is fully happy to use a blank formatted file, it will populate the essentials, just as well as if Puppy made the pup_save file.

Bruce

[Bruce B]

If you having issues with flash, keep trying different ones, until you get one that works best for you pc
ttuuxxx

And if he does, that has nothing to do with Puppy, Linux or anything really but proprietary author.

Flash never had a good reputation for stability, security or privacy. It's not free software. They won't even show the source code, although showing the source doesn't mean free software.

But it does mean that others can't look at it and Macromedia lacks the benefit of a community of brilliant hackers who might be able to see flaws in it, which their 'in house team' doesn't get. The proof they don't get it is in its history.

Main point here is the Macromedia non-free software should be a reflection on Puppy, FOSS or anything other than the Macromedia non-free software.

That is, if the problem can be traced to Macromedia Flash and I don't know that it has been. I'd to see responsibility assigned where it belongs.

Also I'd like to mention resolve this problem.

And honestly, I can leave my Puppy up for days on end. It would actually surprise me if it failed. Occasional application troubles are usually easy to deal with by using various utilities. I use a signal 9 script I wrote or htop to close a bad behaving process.

Bruce

[Flash]

Having skimmed through this thread, now I'm an expert? Anyway, based on what I understood so far it seems to me that blaming NTFS is jumping to conclusions without any supporting proof. Everybody only guessed that the cause of the problems was NTFS.

I sometimes boot my multisession Puppy DVD in a Windows computer that has a hard disk so that I can use Pawdioconverter to convert mp3 CD books to lower resolution mp3 for my mp3 player. Puppy stores the reduced mp3 files on a NTFS partition. I've done this hundreds of times over several years without once losing any files or messing up the NTFS partition. At first, years ago, a few times there was a problem with a folder that showed up in Windows but that Windows said wasn't there and would not delete. After Puppy deleted it everything was ok. This hasn't happened in years.

I run Puppy from a multisession DVD in a computer with 512 MB of RAM but no hard disk drive. For several years after I started running Puppy from a multisession DVD the computer only had 256 MB of RAM. Last summer I discovered that the BIOS could handle 512 MB of RAM if it was in two 256 MB sticks, so I upgraded. I usually turn off the computer at night but I have on occasion left it on for several days. I can't remember ever getting an "out of memory" error or having Puppy freeze on me. Certainly it hasn't happened in several years.

[jrb]

silverojo,
One of the simpler things you can do is remaster. (I know some people don't think that's simple but it gets so with practice)

Get puppy setup the way you like it, install all your favourite .pets and then go to the Setup menu -> Remaster live cd. There are a number of threads on the forum which go into advanced remastering.

The instructions in Remaster aren't too bad. Remaster Express by pa_mcclamrock works really well but is only set up for puppy411 right now.

When you are finished you will have to make a new pup_save to go with your new disk or frugal. If you should crash this pup_save however, you won't have lost any of your custom setup.

If you are careful to save all your personal files to /mnt/home or another partition on your harddrive you won't have lost them either.

[ecomoney]

I agree about the ntfs being buggy when read by linux, interesting stuff about the ntfs format from whodoo. NTFS has to be read write in puppy because users want to get at and change their data while trying puppy. If pppy doesnt work for them theyre likely to stick with buggy, bloated windows forever. Yes its not perfect, but its an improvement.

I was the one who reported to the puppy forum that the ntfs-3g guys had cracked ntfs read-write support back in version 2.03. Since then theyve refined it even more. Does puppy use the latest version?

[ttuuxxx]

[quote="Bruce B"][/quote]
Bruce move your flash then install the latest flash
open a terminal and type
ldd /usr/lib/mozilla/plugins/libflashplayer.so <---if thats the location of your flash
that will show you everything your missing, check out the version numbers like on glibc etc,
ttuuxxx

[amigo]

Resizing a file on NTFS from within Linux could be the cause of corruption -depending on how the read-write setup is done under Linux. Using the native kernel NTFS driver and resizing a file was/is prone to causing the 'dirty' flag to be set on the NTFS partition so that it would need to be 'fixed' by Windows. ntfs-3g does better with this than the native driver, but could well have problems to, depending on the version.

I used to have a distro that was designed primarily to be installed to and run from a disk image on an NTFS drive using the native kernel driver. In order to resize the file, you could run a program under Windows which would cat extra bytes onto the disk image. Then, when rebooting to linux, you could upsize the partition to match the new file size and have the filesystem grow that way.