StretchDog

A home for all kinds of Puppy related projects
Message
Author
User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#221 Post by fredx181 »

Ether wrote:If I don't use EXIT:/... , how does that change the size of ROOT ?
If save file is used you get the amount of space used and total size of the savefile.
If save folder is used you get used and total of the partition the savefolder is on.
Is there a way to store changes on disk in a temporary area, and then let me decide on EXIT if I want to save that session or not?

In other words, I like the functionality of changes=EXIT, but with changes stored in a temporary area on disk, instead of in limited RAM. Is there a boot configuration to do that ?
No such option,sorry, I like the idea though, but not sure if and how it's possible to implement.

Fred

williams2
Posts: 337
Joined: Fri 14 Dec 2018, 22:18

#222 Post by williams2 »

not sure if and how it's possible to implement
It should be very easy to implement.

A standard fugal install with a savefile has the sfs files mounted ro and the savefile mounted rw.

aufs takes these mounted sfs and savefile file systems and mounts them as a layered file system, with the rw layer on top.

Which is exactly what you want. The top filesystem would be empty, of course.

For a usbflash or ataflash file system, Puppy will mount the savefile rw, add it to the aufs layered fs, and add a tmpfs file system as the top layer. So aufs will not write to the savefile, even though it is writable, aufs only writes to the top tmpfs ram layer.

All you need to do is mount the temporary savefile fs as the top layer and the normal savefile as the second layer.

Which should be easy to do.

Also, it should be possible to easily remount the aufs file system any time, while Puppy is running, much the same as adding an sfs file layer. For example, adding a devx sfs. So someone should be able to write a script to do this on the fly, while Puppy is running.

Then you would have the same choices as a usbflash or ataflash has, to copy the changes in the top aufs layer to the savefile layer when you click the desktop icon or be asked if you want to save the changes when Puppy shuts down.

williams2
Posts: 337
Joined: Fri 14 Dec 2018, 22:18

#223 Post by williams2 »

I am running BionicPup64-8.0 in ram with no savefile
with an adrv-sfs.

The amount of space my pup_rw is using is 32MB.

Code: Select all

# du -hs /initrd/mnt/tmpfs/pup_rw
32M	/initrd/mnt/tmpfs/pup_rw
#

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#224 Post by fredx181 »

Hi williams2,
Thanks for the info, but there may be a misunderstanding
aufs takes these mounted sfs and savefile file systems and mounts them as a layered file system, with the rw layer on top.

Which is exactly what you want. The top filesystem would be empty, of course.
I think similar is already the case, "/mnt/live/memory/images/changes-exit" is comparable with "dev_save" in puppy and "/mnt/live/memory/changes" is comparable with "pup_rw"
(or maybe I misunderstood what you mean :?: )

Code: Select all

root@live:~# df -h '/mnt/live/memory/images/changes-exit'
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda7       163G  141G   14G  92% /mnt/live/memory/images/changes-exit
root@live:~# df -h /mnt/live/memory/changes
Filesystem      Size  Used Avail Use% Mounted on
tmpfs           1.2G   50M  1.2G   5% /mnt/live/memory/changes
root@live:~# df -h /
Filesystem      Size  Used Avail Use% Mounted on
aufs            1.2G   53M  1.2G   5% /
The issue is that tmpfs has limited size (same as pup_rw in puppy with pupmode 13)
So trying to find a way to mount /mnt/live/memory/changes on an actual filesystem (in my case /dev/sda7), (so then in my case I got the 14G space left in / , rather than1.2G ) and still have the option to save-on-exit only.
But I got ideas already how to modify the init script to accomplish that.

@Ether
If you are interested to test a modified initrd1.xz, just tell me.
Before that I need to know what architecture and kernel you're booting, the output of "uname -a" will tell me that.
EDIT: and if you use savefile or savefolder.

Fred

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

#225 Post by Ether »

fredx181 wrote:@Ether
If you are interested to test a modified initrd1.xz, just tell me.
Yes.
Before that I need to know what architecture and kernel you're booting, the output of "uname -a" will tell me that.
root@live:/mnt/sda2# uname -a
Linux live 4.9.0-3-686-pae #1 SMP Debian 4.9.30-2+deb9u5 (2017-09-19) i686 GNU/Linux
EDIT: and if you use savefile or savefolder.
save folder.

Thanks.

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

"changes" folder pruning

#226 Post by Ether »

.
I wrote a simple AWK script to identify all files larger than 5M in my changes folder. Here's the report it generated :
8781907 hwdb.bin
./lib/udev:

6962372 goog-phish-proto.vlpset
./root/.cache/mozilla/firefox/s7gs9vjs.default-esr/safebrowsing/google4:

7320070 scriptCache-current.bin
5506672 startupCache.4.little
./root/.cache/mozilla/firefox/s7gs9vjs.default-esr/startupCache:

5242880 favicons.sqlite
5242880 places.sqlite
./root/.mozilla/firefox/s7gs9vjs.default-esr:

120231300 libxul.so
18418940 omni.ja
./usr/lib/firefox-esr:

45192674 omni.ja
./usr/lib/firefox-esr/browser:

13348992 libavcodec.so.57.64.101
./usr/lib/i386-linux-gnu:

27225216 pkgcache.bin
27156073 srcpkgcache.bin
./var/cache/apt:

51046304 firefox-esr_68.4.1esr-1~deb9u1_i386.deb
./var/cache/apt/archives:

38737432 deb.debian.org_debian_dists_stretch_main_binary-i386_Packages
26451928 deb.debian.org_debian_dists_stretch_main_i18n_Translation-en
./var/lib/apt/lists:

total=406865640

Which of the above files can be safely deleted ?


.

dancytron
Posts: 1519
Joined: Wed 18 Jul 2012, 19:20

#227 Post by dancytron »

"27225216 pkgcache.bin
27156073 srcpkgcache.bin "

These 2 are the caches for synaptic/apt-get and can be deleted. I put them as "custom files" in bleachbit so they get deleted whenever I run it. They will be instantly recreated, but at a smaller size. I usually have to run bleachbit 2x to get them to be their smallest.

"51046304 firefox-esr_68.4.1esr-1~deb9u1_i386.deb
./var/cache/apt/archives: " Is a cached .deb file for something you installed. Bleachbit or running apt-get clean or apt-get autoclean should get rid of it automatically, but you can just delete it if you want.

If you care about this kind of thing, I recommend you try bleachbit.

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

#228 Post by Ether »

.
dancytron wrote:"27225216 pkgcache.bin
27156073 srcpkgcache.bin "

These 2 are the caches for synaptic/apt-get and can be deleted. I put them as "custom files" in bleachbit so they get deleted whenever I run it. They will be instantly recreated, but at a smaller size. I usually have to run bleachbit 2x to get them to be their smallest.

"51046304 firefox-esr_68.4.1esr-1~deb9u1_i386.deb
./var/cache/apt/archives: " Is a cached .deb file for something you installed. Bleachbit or running apt-get clean or apt-get autoclean should get rid of it automatically, but you can just delete it if you want.
Thanks.
If you care about this kind of thing, I recommend you try bleachbit.
I will Google that and read the user manual.
.

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#229 Post by fredx181 »

Hi Ether,

Here's modified initrd1.xz for kernel 4.0.9-3-686-pae 32 bit
https://dl.dropboxusercontent.com/s/dym ... BEXIT?dl=1
Rename to initrd1.xz and replace in "live" (better backup original first)
Added option "+EXIT", usage for example on the kernel boot line:

Code: Select all

changes=+EXIT:/stretch_dog/live
will create folder "tmp-changes" in live and will be written to the save storage on demand (e.g. by running save2flash or choosing Save at shutdown).
Note that it's not a tmpfs, changes are not written to RAM (as is the case with normal EXIT), so speed will be slower.

Fred

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

#230 Post by Ether »

fredx181 wrote:Hi Ether,

Here's modified initrd1.xz for kernel 4.0.9-3-686-pae 32 bit
https://dl.dropboxusercontent.com/s/dym ... BEXIT?dl=1
Rename to initrd1.xz and replace in "live" (better backup original first)
Added option "+EXIT", usage for example on the kernel boot line:

Code: Select all

changes=+EXIT:/stretch_dog/live
will create folder "tmp-changes" in live and will be written to the save storage on demand (e.g. by running save2flash or choosing Save at shutdown).
Note that it's not a tmpfs, changes are not written to RAM (as is the case with normal EXIT), so speed will be slower.

Fred
It booted successfully and there's a "tmp-changes" folder in the "live" folder.

I'm posting this response while booted to that configuration.

If I encounter any issues or questions related to use of this configuration I'll share them here.


.
Attachments
2020-02-03-161044_374x278_scrot.png
(23.34 KiB) Downloaded 376 times
2020-02-03-161029_206x137_scrot.png
(19.08 KiB) Downloaded 376 times
2020-02-03-165425_398x402_scrot.png
(42.91 KiB) Downloaded 377 times

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

New option "+EXIT" added for porteus-boot

#231 Post by fredx181 »

*** New option "+EXIT" added for porteus-boot ***
(save on exit or save on demand, modified)

In case anyone interested to try.
See above (and previous page) for more info about the subject, in short:
When using "+EXIT" the changes will NOT run in memory, which can be an advantage, specially when you have small amount of RAM.
(with the "normal" EXIT: configuration, the amount of space in / is limited to 60% (default) of the amount of RAM you have).

To be able to use "+EXIT", a new initrd1.xz must be created, therefore:
- Install latest "upgrade-kernel2" v0.0.3
https://fredx181.github.io/StretchDog/a ... .3_all.deb
- Run from menu > System > Make initrd (see pic below)
- When done, copy /tmp/initrd1.xz to your "live" frugal install folder and replace the existing initrd1.xz
- Reboot and use for example on the kernel boot line:

Code: Select all

changes=+EXIT:/stretch_dog/live
EDIT: Note that this option can only be used when "live" is on a Linux filesystem (e.g. ext3 or ext4)
EDIT2: Corrected above link to upgrade-kernel2

P.S. Upgrade-kernel v2 needed an update anyway (previous version didn't work anymore)
Run from Menu > System > Upgrade Kernel v2 to upgrade the kernel to latest (currently 4.9.0-11)
When done there should be folder containing e.g. k-4.9.0-11-i386-pae.squashfs, vmlinuz, initrd.img, initrd1.xz, these can be copied to the "live" folder to boot with version 4.9.0-11)
(btw, the initrd1.xz created will also have the new option +EXIT included)

Fred
Attachments
01-mkinitrd.png
Make initrd
(77.13 KiB) Downloaded 312 times
Last edited by fredx181 on Fri 07 Feb 2020, 21:06, edited 1 time in total.

backi
Posts: 1922
Joined: Sun 27 Feb 2011, 22:00
Location: GERMANY

#232 Post by backi »

I don`t really need it .......but applause to Fred...... another nifty Gadget from "Gyro Gearloose- The Madcap Inventor" ....a cool Expansion for users with low Ram .......
just cool ......!
Attachments
Gyro Gyro Gearloose.jpg
(120.46 KiB) Downloaded 280 times

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

#233 Post by Ether »

backi wrote:I don`t really need it .......but applause to Fred...... another nifty Gadget from "Gyro Gearloose- The Madcap Inventor" ....a cool Expansion for users with low Ram .......
just cool ......!
I've been using it for over a week now of long browsing sessions with no crashes.

I do have a question though.

Conky says root is 4.3GB
YAD (How Bigs?) says 464MB
properties says 485MB

What is each one actually reporting ?
Attachments
2020-02-12-155023_234x220_scrot.png
(11.48 KiB) Downloaded 230 times
2020-02-12-155031_203x139_scrot.png
(18.97 KiB) Downloaded 233 times
2020-02-12-155102_424x464_scrot.png
(38.17 KiB) Downloaded 235 times

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#234 Post by fredx181 »

Ether wrote:Conky says root is 4.3GB
That's not about the 'root' directory.

Conky: 4.29GIB/12.8GIB means that on your 'changes' partition there's 4.29GIB used from a total of 12.8GIB

You can check it out too with df -h /mnt/live/memory/changes (with +EXIT the "tmp-changes" folder is a bind mount to /mnt/live/memory/changes), my output:

Code: Select all

root@live:~# df -h /mnt/live/memory/changes
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda7       163G  141G   15G  91% /mnt/live/memory/changes
My partition sda7, used: 141G, total: 163G and conky says: 141GiB/163GiB

The size of directory "root" is another thing, "How Big?" and Thunar have different ways of counting, I guess. (I personally trust How Big? better)

Fred

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

#235 Post by Ether »

.
Well after more than 2 weeks using Fred's StretchDog32 with intense browsing, I've discovered a serious issue.

Google Maps "Street View" loads very slowly, then does not respond to attempts to click-and-drag the view.

Street View loads very quickly in WinXP. Click-and-dragging the view works smoothly and responsively. So it's not a hardware issue.

Do I need a non-free video driver of some sort?

This machine has video controller integrated on the motherboard.

What is the Linux command to display info about the video controller?

.

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

#236 Post by Ether »

Ether wrote:.
Well after more than 2 weeks using Fred's StretchDog32 with intense browsing, I've discovered a serious issue.

Google Maps "Street View" loads very slowly, then does not respond to attempts to click-and-drag the view.

Street View loads very quickly in WinXP. Click-and-dragging the view works smoothly and responsively. So it's not a hardware issue.

Do I need a non-free video driver of some sort?

This machine has video controller integrated on the motherboard.

What is the Linux command to display info about the video controller?

.
I booted to TahrPup which has Hardinfo installed
and looked up the video controller hardware :

Intel 82945G/GZ (rev02)(prog-if 00 [VGA controller])


.

dancytron
Posts: 1519
Joined: Wed 18 Jul 2012, 19:20

#237 Post by dancytron »

Ether wrote:.
Well after more than 2 weeks using Fred's StretchDog32 with intense browsing, I've discovered a serious issue.

Google Maps "Street View" loads very slowly, then does not respond to attempts to click-and-drag the view.

Street View loads very quickly in WinXP. Click-and-dragging the view works smoothly and responsively. So it's not a hardware issue.

Do I need a non-free video driver of some sort?

This machine has video controller integrated on the motherboard.

What is the Linux command to display info about the video controller?

.
Sounds like a browser issue.

First thing I'd do is try deleting the cache.

What browser are you using?

Whatever it is, get a different one (one of the portable browsers) and see if that helps.

I don't think it is a graphics card issue, but install mesa-utils and then from a terminal window run "glxgears -info". Do the gears work? What does it say about what graphics driver is running?

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#238 Post by fredx181 »

Also it may help to install "libgl1-mesa-dri" .

Fred

Ether
Posts: 261
Joined: Wed 21 Aug 2013, 17:56

#239 Post by Ether »

dancytron wrote:
Ether wrote:.
Well after more than 2 weeks using Fred's StretchDog32 with intense browsing, I've discovered a serious issue.

Google Maps "Street View" loads very slowly, then does not respond to attempts to click-and-drag the view.

Street View loads very quickly in WinXP. Click-and-dragging the view works smoothly and responsively. So it's not a hardware issue.

Do I need a non-free video driver of some sort?

This machine has video controller integrated on the motherboard.

What is the Linux command to display info about the video controller?
.
Sounds like a browser issue.

First thing I'd do is try deleting the cache.

What browser are you using?

Whatever it is, get a different one (one of the portable browsers) and see if that helps.

I don't think it is a graphics card issue, but install mesa-utils and then from a terminal window run "glxgears -info". Do the gears work? What does it say about what graphics driver is running?
Here's what I've found so far:

Firefox 68.5.0esr-1~deb9u1 is the one that has the Google Street View click-and-drag problem.

So I tried Chromium 73.0.3683.75-1~deb9u1 and it works like a dream: it loads Street View immediately, and click-drag works smoothly and responsively.

Problem is, Chromium won't run when booted to root unless I use the --no-sandbox option (which gives a security warning).

And if I Exit Xterm and login as puppy, it won't let me back into Xterm (permission denied error).


.

dancytron
Posts: 1519
Joined: Wed 18 Jul 2012, 19:20

#240 Post by dancytron »

try to fix firefox with this. https://support.mozilla.org/en-US/kb/re ... d-settings

if that doesn't work, try to uninstall and reinstall

Post Reply