antiX live usb issue

[BitJam]

Thanks guys! Sorry for the problem. I'll look into it.

[anticapitalista]

rs55 - please try what I posted further up ie bugfix because I had the exact same vmlinuz issue and the fix 'fixed' it.

bump bump bump

[rs55]

rs55 - please try what I posted further up ie bugfix because I had the exact same vmlinuz issue and the fix 'fixed' it.

bump bump bump

Sorry - I ve been trying so many things - losing track! Yes installing fuseiso9660 ( as you suggested) does fix the problem. But as discussed earlier in this thread, thats only because lum is looking for fuseiso , and when it does'nt find it, it does'nt run it - and so neither fuseiso or fuseiso9660 were getting run. So it worked.
I dont have an immediate issue - because simply uninstalling fuseiso does the trick. But I think BitJam is trying to figure out how to have lum run fuseiso and still work on my machine( because many other users seem to need fuseiso for lum to work).

[anticapitalista]

Did you install the updated live-usb-maker-gui-antix db from the dev repo as I posted? If so, does the gui app now produce a bootable live usb? Same for cli lum?

If so, then the problem (as I have said many times) is the existence of fuseiso. (not fuseiso9660).

Therefore, we (Bitjam) need to remove fuseiso from our scripts and use fuseiso9660 instead.

[rs55]

Did you install the updated live-usb-maker-gui-antix db from the dev repo as I posted? If so, does the gui app now produce a bootable live usb? Same for cli lum?

If so, then the problem (as I have said many times) is the existence of fuseiso. (not fuseiso9660).

Therefore, we (Bitjam) need to remove fuseiso from our scripts and use fuseiso9660 instead.

oh - I did'nt try your gui. I will. yesterday, I had installed fuseiso9660 and changed a line in the cli-shell-utils.bash to local prog = fuseiso9660. That failed. I dont know if your gui app essentially does that - or something else. I'll try it and report back.

[rs55]

On antiX, add the word dev to the /etc/apt/sources.list.d/antix.list so it reads

deb http://repo.antixlinux.com/stretch stretch main nosystemd nonfree dev

Code: Select all

apt-get update && apt-get dist-upgrade

This will install a bugfix version of lum (gui) and force removal of fuseiso (if installed) and it will install fuseiso9660

This creates a bootable live usb on my antiX-17 box.

anticapitalista:
I got your gui package from dev. It ran great. produced a bootable usb. And I checked that fuseiso9660 continued to be installed ( I had already manually installed it earlier). Great! Nice work.
Hope this - or some variant of it makes it into the stable channel down the road.

[rs55]

actually - this fix needs to go into the stable channel with some urgency. because the current stable download fails - at least on my system. That is a pretty big bug - on the most amazing feature of antiX - the live-usb system.

[rs55]

Can you tell us what that gui does? is it calling the lum --force=nofuse ? Well, if you are actually installing fudeiso9660, it would'nt make much sense to then have it not called - I suppose !!
Quite curious how you fixed it.

[BitJam]

I was unable to replicate the "could not find syslinux or isolinux directory" problem running on a live version of antiX-17.4. I also didn't have the problem on my development machine when I ran from the unpacked tarball. I tried it again inside an X-windows terminal in antiX-17.4 and it still worked even though the efi partition and sometimes the main partition got automounted while LUM was running.

If you are running in antiX, try using the --force=automount option to disable antiX automounting while the program runs.

I have no idea why the changes I made would affect the part of the code where you got the error message but it could be a timing issue that lets automount "steal" my partitions. While I was running LUM, in another terminal window I also ran:

Code: Select all

watch -n .2 df -Th

to see what was mounted and what was not. It appeared that without --force=automount, the mountpoint at /run/live-usb-maker/main could disappear which could cause a problem if I'm trying to install the bootloader.

ISTM that in addition to being over-aggressive, the automounting is broken. I would often end up with mountpoints at both /media/antiX-uefi and /media-antiX-uefi-2 for the same partition.

If automounting is causing the problem I could fix it on antiX by reversing the sense of --force=automount but that won't fix it on other systems such as MX. But maybe it is not a problem on other system.

Another solution would be to add delays to give the automount programs a chance to steal my partition and then I could --bind mount from there. I could also try to stop using partprobe after I've changed the partitioning but that can confuse the system and prevent *me* from mounting the new partitions.

[BitJam]

Once more unto the breach, dear friends, once more

This is the same as before but I changed the sense of --force=automount so automounting is disabled by default on antiX. I also added a 2 second delay (*sigh*) between when I create the new partitions and when I grab them back from the automounter. You could test this by using --force=automount which will now allow antiX automounting while we are running.

But the first priority is to see if this new code fixes the fragmentation problem on rs55's system when fuseiso and/or fuseiso9660 is used.