Saluki

[jemimah]

Well... /tmp/bootcnt.txt can be ...

deleted in rc.shutdown .. in full installs .. or frugal pupmode 12 on HDD
also unmounted in rc.shutdown for frugals having /tmp mounted as tmpfs ..

Barry has done some code adjustments to rc.shutdown recently .

Is Saluki still based on racy woof 5.2(.x) ?

(One note about thunar : Thunar uses a trash dir , so i was having problems deleting files larger than the remaining free space in the pupsave-file)

Yes still based on 5.2. But rc.shutdown from pupsaveconfig is newer. I can roll back to the old pupsaveconfig or we can try to fix the newer rc.shutdown.

I had the occasional "session not saved" problem with the older pupsaveconfig - so it'd be nice if the new version could be made to work.

[shinobar]

I can roll back to the old pupsaveconfig or we can try to fix the newer rc.shutdown.

Problem is at the rc.sysinit.

if [ ! -d /initrd ];then #w468
busybox mount -o remount,rw /
#120409 no longer deleting /tmp/* in rc.shutdown... (note, init script in initrd.gz wipes it)
rm -rf /tmp/*
rm -rf /tmp/.[0-9a-zA-Z]*
echo 'PUPMODE=2' > /etc/rc.d/PUPSTATE
if [ "$ORIGLANG1" != "en" ];then #120217
echo "OUTPUT_CHARSET=UTF-8
export OUTPUT_CHARSET" >> /etc/rc.d/PUPSTATE
fi
fi
. /etc/rc.d/PUPSTATE #variables created at bootup by init script in initrd.

No need to change the rc.shutdown, but the deletimg /tmp/* is required in the rc.sysinit.

[jemimah]

I can roll back to the old pupsaveconfig or we can try to fix the newer rc.shutdown.

Problem is at the rc.sysinit.

if [ ! -d /initrd ];then #w468
busybox mount -o remount,rw /
#120409 no longer deleting /tmp/* in rc.shutdown... (note, init script in initrd.gz wipes it)
rm -rf /tmp/*
rm -rf /tmp/.[0-9a-zA-Z]*
echo 'PUPMODE=2' > /etc/rc.d/PUPSTATE
if [ "$ORIGLANG1" != "en" ];then #120217
echo "OUTPUT_CHARSET=UTF-8
export OUTPUT_CHARSET" >> /etc/rc.d/PUPSTATE
fi
fi
. /etc/rc.d/PUPSTATE #variables created at bootup by init script in initrd.

I can't sync with woof right now. Rolling back pupsaveconfig is a safer fix. Isn't the goal of pupsaveconfig to work with older versions of woof? If not, it would be good to put a note in the thread so devs to get the right version for their puplet.

[shinobar]

Isn't the goal of pupsaveconfig to work with older versions of woof?

Yes, pupsaveconfig works with old wooof but also with the old woof bug.

If the finding of the moving 'rf -f /tmp*' from shutdown to the init is real solution, we are better to move. (but i am not quite sure what is the real cause.)

[jemimah]

That one change is fine but I'm likely to introduce a lot of new problems by only partially syncing. I'm not convinced the rm -rf thing is the issue though. Having it in rc.shutdown worked fine in the past and should continue to work.

If I could reproduce the issue I could probably debug it.

[shinobar]

If I could reproduce the issue I could probably debug it.

So do i. i cannot reproduce it and still am not sure.

[l2ulinux]

I have been using Saluki from the start. By doing a frugal install with only a few problems. Found that it was easier to delete the older install and make a total new one.
Only once did I have a problem with it not shutting down after a frugal install. All it did was boot back to the desktop and when I wanted to shut down it did with no problem.
Now I have it install on four system with each a different setup. No problem with any and love to use it.

[Karl Godt]

Reading the manpage of mingetty i found that it seems to need access to /var[/run/utmp] .

Seems that /var is by default

drwxr-xr-x and owned by root:root

i am not running Puppy in other USER mode than root ...
except few times to try to reproduce sound issues posted on the forum which seems to have been fixed by a udev rule to chown /dev directory ..
(i did it with /etc/init.d/alsa) ..

[mikeslr]

Hi jemimah & all;

This is not a vital post. Just reporting a curiosity.
I was exploring playdaz's latest iteration of Lupu, scrolling thru the pets I've stored for something else and noticed a Hulu pet I had forgotten I downloaded sometime ago. So I installed and started it in Lupu. I ran, but complained about being out of date: offering to "update." Figuring that could break something, I deleted the Hulu pet, but --then being in Lupu-- decided to download the current "ubuntu" version from ubuntu. That ran without a hitch.
Now, wondering if it would run in Saluki, I booted into Saluki 20, clicked install and then opened Hulu. It complained about being unable to locate Flash, offering the alternatives of your manually editing Hulu's config file (/.huludesktop), or downloading flash.
On that Saluki, I have installed firefox from Saluki's repo, Opera downloaded from Opera to /mnt/home and run via a hand-made pet which created a script in /my-applications/bin and a desktop file, and Google-Chrome-Stable.sfs from Saluki's repo. All played flash files. I had assumed that Opera was employing the same flashplayer.so as firefox, which from prior experience had been installed to /usr/lib/mozilla/plugins, and Google-Chrome was using its own. Uncertain what downloading another flash-player would do to my system, I decided to edit /.huludesktop, opening it in geany.
Scrolling revealed a line, which at the time read “flash_location = (null). As I was uncertain that my memory was accurate regarding the location of the flash-player.so, I opened pfind and entered the argument “flash.

[jemimah]

Flash is not installed unless you installed it from the ppm. If youtube works, it's because it's using html5 video.

[jemimah]

Reading the manpage of mingetty i found that it seems to need access to /var[/run/utmp] .

Seems that /var is by default

drwxr-xr-x and owned by root:root

i am not running Puppy in other USER mode than root ...
except few times to try to reproduce sound issues posted on the forum which seems to have been fixed by a udev rule to chown /dev directory ..
(i did it with /etc/init.d/alsa) ..

Remember that mingetty is not actually running if there is an open shell on tty1. It gets executed by init when the tty1 shell exits. Once mingetty opens a new shell, it exits. So its the shell that may or may not be getting killed - not mingetty.

If the new shell opens while /tmp/bootcnt.txt does not exist, then X starts - otherwise you get a prompt.

[greengeek]

Perhaps the Legacy button should be removed at this point?

I think that would be a good idea unless it is possible to work out why it is buggy. I guess people who are more experienced with usb installs would have already learned to ignore this option. Or does it apply to all "FULL" installs, not just non_save_to_file usb installs?

Well done to the eagle-eyed drblock2!

Sage, which topic is his post in?

[greengeek]

So do i. i cannot reproduce it and still am not sure.

Are you referring to the looping back to desktop problem? If so, here is how I generated the problem repeatedly on my Acer netbook:
Boot Saluki 20 from CD, pfix=ram
Insert blank usb stick
Run universal installer and when it brings up Gparted create 3 partitions: part 1: Fat32 for data, part2: Ext2 "boot", part 3: swap. (no idea if this particular partition config has any effect or not)
Install to partition 2. Accept default mbr, wipe all files. Choose NOT to load files to ram at boot time.
Shutdown, remove CD drive.
Reboot, change to en_AU (two step process as you have to select standard en english first, then select en_AU from the next list that pops up, then allow restart of X. Then shutdown, then accept "legacy" save, and select the usb ext2 partition. (512Mb)
Wait long time till shutdown.

Future reboots/shutdowns should show the looping reboot problem. (although I have seen variable behaviour - sometimes it succeeds on the very first shutdown after locale config, then fails on later shutdowns. Test thoroughly)

EDITED Needed to clarify that this was my test with Saluki 20 only - I am just about to start testing with Saluki 21 next. Also clarified that my chosen locale was Australian english. Added the step to NOT copy files into ram at boot time.

[Karl Godt]

Just a short note about the shutdown problems :

1) It seems that it is an overall problem that should be discussed in a separate general thread .
2) I am responsible to some adjustments lately
and after looking inside the code i can only guess :

2a) regular sed depends on locale settings .. especially LC_ALL and as stated in the html of the source that might not compile because the devx might not include many text formatting binaries LC_COLLATE and LC_CTYPE

2aa) LC_COLLATE and LANG are defined in /etc/profile (login,*getty) and should get adjusted by the various Puppy locale setting scripts (chooselocale,firstrunwizard,...)

2ab) I don't actually have a mount output example for PUPMODE=6 (save to entire partition) .. it might be possible that the home partition gets unmounted or remounted read-only too early

2ac) busybox can be configured with locale(s) or utf-8 option(s) -> init,umount

2b) the killzombie function of rc.shutdown might catch a defunct process and might kill a process that should better not be killed ??? (ps -A -H in urxvt could reveal them before selecting a shutdown button in the menu)

2c) I have no idea about gettext and eval_gettext .. ??

2d) rc.shutown attempts to killall X (another time) .. ??

2e) the whole LAN mount things are unknown to me, so i don't know if bringing down the network attached to local routers could be responsible for anything (especially mounted puppy archive files (.sfs,.2fs) on network shares)

[greengeek]

Just in case it helps to narrow down the problem I thought I would mention the following:

If I install Gray's Racy 5.2.2 NOP in the same way as I have been installing Saluki 20 I do not have any shutdown problems at all.

NOP 5.2.2 seems to use a different routine for specifying the locale - I can choose en_AU on the first stage - it does not require me to go to a second screen to select en_AU, Also, it does not offer the "Legacy" prompt to do the install to partition. I have now tried both the "save to file" and "save to partition" with NOP 5.2.2 and they both work fine.

EDITED have corrected spelling of aus locale to en_AU to reflect choice of Australian english.
.

[greengeek]

I would also like to mention another problem I am having with Saluki, as it might remotely be related to these shutdown issues (but probably not). Again, I have this problem with Saluki, but not with NOP 5.2.2 which is based on the same racy.

DISCLAIMER: this will sound a bit spooky-tunes. You will think I am going mad. Maybe I am. Anyway, here is my problem:

When I try to boot from the usb stick I have loaded Saluki onto (regardless of whether I selected "save to file" or "save to partition") the system will not boot on the first attempt. It will hang forever showing:

SYSLINUX 4.05 EDD 0x4f131b2f Copyright (C) 1994-2011 H Peter Anvin et al.

If I force a poweroff by holding the power button for 10 seconds, then power on again within 20 seconds, Saluki will boot up correctly. If I leave it longer than 20 seconds, the next power on will go back to the SYSLINUX prompt and hang there forever. (actually - it is not a prompt, it is just like the machine stopped in mid sentence)

I also notice that if I remove and reinsert the usb stick during the initial 20 seconds after a poweroff, it will not reboot - just goes to SYSLINUX message again. This suggests to me that perhaps my machine maintains some portion of usb power for approx 20 seconds after poweroff? And that this affects whether or not Saluki boots correctly.

Or maybe the netbook power management system has something that still exists in memory or BIOS for 20 seconds after a power off? I just don't know. Anyway, it is repeatable and happens every time, and I have tried reinstalling about 10 times on two (identical) usb sticks, and now a third, different, higher quality stick, without being able to get rid of the problem. NOP 5.2.2 does not show this behaviour.

[Karl Godt]

Just in case it helps to narrow down the problem I thought I would mention the following:

If I install Gray's Racy 5.2.2 NOP in the same way as I have been installing Saluki, I do not have any shutdown problems at all.

NOP 5.2.2 seems to use a different routine for specifying the locale - I can choose en_Aus on the first stage - it does not require me to go to a second screen to select en-Aus, Also, it does not offer the "Legacy" prompt ro do the install to partition. I have now tried both the "save to file" and "save to partition" with NOP 5.2.2 and they both work fine.

Just to be precise : I think you refer to en_AU because you are located in NZ (en_NZ) OR en_US ?

Could you please post the output of the

Code: Select all

locale

Code: Select all

locale -a

Code: Select all

locale -m

commands ?

[Brown Mouse]

When I try to boot from the usb stick I have loaded Saluki onto (regardless of whether I selected "save to file" or "save to partition") the system will not boot on the first attempt. It will hang forever showing:

SYSLINUX 4.05 EDD 0x4f131b2f Copyright (C) 1994-2011 H Peter Anvin et al.

I see the same thing when booting Lucid 528-05 from an sdhc on my Acer Aspire one netbook.

[Geoffrey]

(One note about thunar : Thunar uses a trash dir , so i was having problems deleting files larger than the remaining free space in the pupsave-file)

Yes, I've had that problem with a usb install also, if a file is deleted in /mnt/home then it creates /mnt/home/.Trash-0 which can not be emptied if there isn't sufficient free space,
this results in having to reformat the drive, I now don't delete files in /mnt/home
It might be an idea to be able to select a percentage of free space for the trash like that other OS.

[Karl Godt]

(One note about thunar : Thunar uses a trash dir , so i was having problems deleting files larger than the remaining free space in the pupsave-file)

Yes, I've had that problem with a usb install also, if a file is deleted in /mnt/home then it creates /mnt/home/.Trash-0 which can not be emptied if there isn't sufficient free space,
this results in having to reformat the drive, I now don't delete files in /mnt/home
It might be an idea to be able to select a percentage of free space for the trash like that other OS.

I finally opened a terminal in these trash directories and did a

Code: Select all

rm *
rm .*
cd ..
rmdir *
rmdir .*

to solve this manually