Refracta Development, Scripts, etc.
Post a reply

non-matching md5? please advise

Tue May 31, 2016 9:45 pm

Code:
cat refracta8_xfce_i386_beta-20160402_0213.iso.md5 ; md5sum refracta8_xfce_i386_beta-20160402_0213.iso ; du -b refracta8_xfce_i386_beta-20160402_0213.iso

e204d92c28704595d2ba147908277f79  refracta8_xfce_i386_beta-20160402_0213.iso
676d5dca924331c6e2d291684cec2bda  refracta8_xfce_i386_beta-20160402_0213.iso

206338338   refracta8_xfce_i386_beta-20160402_0213.iso


downloaded from sourceforge (refracta testing)

Re: non-matching md5? please advise

Wed Jun 01, 2016 12:34 am

Code:
e204d92c28704595d2ba147908277f79
is correct. That's what I get on my copy, but that's not what I get with the copy on the sf server.

No, it's not wget. ****! Throw that one away. I removed it from the server. The amd64 iso matched the md5 file, but I don't have the original to compare. Must have deleted it by accident. I've got over 100 snapshots in my jessie folder. Need to get rid of some more.

I'm uploading a newer i386 iso, and I'll replace the amd64, too. In addition to the md5 file, the isos are now gpg-signed, so there a .asc file with it.

Code:
gpg --verify isofile.iso.asc


You might need to get my public key for that to work.
Code:
gpg --recv-keys acdcbe05

Re: non-matching md5? please advise

Wed Jun 01, 2016 12:45 am

No, it's not wget. Fuck! Throw that one away. I removed it from the server.


ahh :) dont feel bad, i found a similar problem with files.devuan.org -- whichever iso is the least popular is the most likely culprit.

in the meantime im trying this:

Code:
cat unofficial_live_devuan_beta_i386_snapshot-20160506_1100.iso.md5\?r\=\&ts\=1464739646\&use_mirror\=netix ; md5sum unofficial_live_devuan_beta_i386_snapshot-20160506_1100.iso
2f4eb4b5f9647216aaeadfa83664d5cb  unofficial_live_devuan_beta_i386_snapshot-20160506_1100.iso
2f4eb4b5f9647216aaeadfa83664d5cb  unofficial_live_devuan_beta_i386_snapshot-20160506_1100.iso


i will definitely try "refracta proper" again, but for now the devuan live is booting and running nicely. cheers.

progress!

Tue Jun 07, 2016 6:03 am

Code:
cat refracta8_xfce_i386_beta-20160526_1442.iso.md5 ;  md5sum  refracta8_xfce_i386_beta-20160526_1442.iso
0c7ee3d92bff59f805aa55bc9c1aa6b0  refracta8_xfce_i386_beta-20160526_1442.iso
0c7ee3d92bff59f805aa55bc9c1aa6b0  refracta8_xfce_i386_beta-20160526_1442.iso


its really great to have load to ram again-- and from the menu even! nice distro so far.

will respect if the laws been laid down on this matter, however is there a shortcut for running the term? ctrl-alt-t or super-t are fine defaults. (former is from some window managers, while the latter is typical for "modern" desktops. i try both when i use a new distro. (if i install icewm i dont have to set it. but it loads whatever is /usr/bin/xterm.)

also loving the focus on mouseover, very leet (its actually my preference, though i find it useless except for heavy term use-- which its great for-- and almost never set it up.)

another great distro. out of curiosity fsr, does your /var/lib/dbus/machine-id start with b4da? im curious if this is a fixed file (and if it can be deleted in the next version.) if its fixed/ro, it is definitely false info-- and i would be happy to see that freedesktop.org feature go. never was a coppermine fan: https://en.wikipedia.org/wiki/Pentium_I ... acy_issues (devuan hasnt fixed this yet either, but deleting it seems to be fine so far.)

Re: non-matching md5? please advise

Wed Jun 08, 2016 2:04 am

Glad you like it. I fell in love with 'focus under mouse' when I was using fvwm2 in suse (with 250mb ram).

The machine-id is excluded from the rsync copy, so whatever you see was generated on your machine. Mine is definitely a different number.

I don't know the key combination to get a terminal. In the wheezy version of refracta, you could do alt-F2, and the Run dialog would come up with xfce4-terminal already in the text-entry box. But that's now a search box. You can still do that, and type it in, in case your mouse doesn't work when you boot up.

Applications Menu -> Settings -> Keyboard -> Application Shortcuts -> Add -> xfce4-terminal -> OK
then press Super-t when the next box comes up. I don't know if that will work right in virtualbox.

Re: non-matching md5? please advise

Thu Jun 09, 2016 1:35 am

The machine-id is excluded from the rsync copy, so whatever you see was generated on your machine.


i read it was generated by install, but this is live. perhaps we could chattr +i that thing instead of excluding it, i dont know. certainly there is nothing to stop a checksum from being generated if someone installs a utility to create one, but i find this default (which i hope will also be taken care of in devuan) very not-right: http://www.freedesktop.org/software/sys ... setup.html

certainly i can chattr +i a file, the question is whats best for refracta.

i wouldnt usually be pushy about that. but this is something i will ultimately complain about in devuan too, if they dont fix it. definitely not picking on refracta here-- though i think refracta could fix it faster (im happy to keep troubleshooting it if that will help.)

Re: non-matching md5? please advise

Thu Jun 09, 2016 12:01 pm

Sorry, I don't understand what you want to do with the machine-id. In the case of /var/lib/dbus/machine-id in a live, read-only system (live-CD), the fact that it's not present in the iso results in getting a new number every time you boot. I don't know if anything needs to use that number.

In the case of systemd's etc/machine-id, I borrowed the following code from live-build. I don't have a current systemd iso to test whether or not you get the same number every boot.
Code:
# Blank out systemd machine id. If it does not exist, systemd-journald
# will fail, but if it exists and is empty, systemd will automatically
# set up a new unique ID.

if [ -e "$work_dir"/myfs(/)etc/machine-id ]
then
   rm -f "$work_dir"/myfs(/)etc/machine-id
   : > "$work_dir"/myfs/()etc/machine-id
fi

Re: non-matching md5? please advise

Thu Jun 09, 2016 12:15 pm

fsmithred wrote:the fact that it's not present in the iso results in getting a new number every time you boot. I don't know if anything needs to use that number.


its a feature of freedesktop.org (afaik, systemd-related) that creates some kind of checksum based on something to do with the machine.

i *really* dont think it should be generated in the first place. this is something devuan should fix-- but if its trivial to fix in refracta first, lets not wait.

deleting it on boot suffices, so long as it has the chance to generate and doesnt get generated "again" later.

again, this is a trivial user-fix after installation (only if you think of it on every install,) but imo you shouldnt be finding a viable machine-id if youre running live -- please lets cut off this "feature" if we can, and then perhaps convince devuan to do the same.

---

if this line is already part of refracta:

Code:
   : > "$work_dir"/myfs/()etc/machine-id


and despite that, a machine-id still gets generated, then this (additional) fix might work:

Code:
   : > "$work_dir"/myfs/()etc/machine-id ; chattr +i "$work_dir"/myfs/()etc/machine-id


im not calling it elegant; i dont want a checksum there, i cant imagine what person against systemd would want a systemd machine checksum by default.

im also surprised that the "thing" creating this id is still there, wherever it is.

Re: non-matching md5? please advise

Thu Jun 09, 2016 2:48 pm

Just leaving myself some notes for later:

Code:
grep -ir machine-id /usr/lib/dbus-1.0/
Binary file /usr/lib/dbus-1.0/dbus-daemon-launch-helper matches

locate machine-id
/usr/share/live/build/hooks/0120-remove-dbus-machine-id.hook.chroot
/usr/share/live/build/hooks/0180-remove-systemd-machine-id.hook.chroot
/var/lib/dbus/machine-id

Re: non-matching md5? please advise

Thu Jun 09, 2016 8:25 pm

figlfdev, presence of the machine-id string chafes me too. IIRC, debian began injecting that prior to the systemd changeover. A daemon (sth like "uuidd") is also autostarted; reputedly the daemon was adopted toward "hardening" the O/S by enhancing random number generation... yet the only thing I could find (after installing gobs of apps) utilizing the daemon was the script generating the machine-id. My takeaway is/was that it represents a self-serving mechanism for the debian devs (facilitating "popcon" install counts).
Post a reply