shibboleth | anyone care to ELI5 the reason for machine-id even existing? | 00:04 |
---|---|---|
shibboleth | management? | 00:04 |
KatolaZ | shibboleth: ask systemd developers :) | 00:04 |
KatolaZ | and BTW a hostid has always existed in unix and unix-like OSes | 00:05 |
shibboleth | asking systemd devs is a lot like asking big tobacco i reckon | 00:06 |
shibboleth | "why is this, exactly"? "don't worry about that, these are not the droids you are looking for, look, a squirrel!" | 00:07 |
KatolaZ | shibboleth: then ask yourself :D | 00:07 |
MinceR | 09 000515 < KatolaZ> and BTW a hostid has always existed in unix and unix-like OSes | 00:08 |
MinceR | hostname? | 00:08 |
KatolaZ | again, /etc/hostid exists since 1982 | 00:08 |
KatolaZ | nope MinceR | 00:08 |
KatolaZ | /etc/hostid | 00:08 |
KatolaZ | it's even in POSIX | 00:09 |
KatolaZ | ans in SUS | 00:09 |
KatolaZ | s/ans/and | 00:09 |
MinceR | looks like this slackware install violates posix | 00:09 |
KatolaZ | maybe :) | 00:09 |
KatolaZ | to be precise, POSIX defines gethostid(2) | 00:11 |
KatolaZ | and it should return a unique, consistent 32-bit identifier of the current host | 00:11 |
KatolaZ | in most implementations, this is kept in /etc/hostid | 00:11 |
MinceR | skeletons keep falling out of the closet | 00:12 |
KatolaZ | or in /var/adm/hostid | 00:12 |
KatolaZ | MinceR: there is no skeleton there | 00:12 |
KatolaZ | it has been there since 4.2BSD | 00:12 |
KatolaZ | we are talking about an "old-school unix" from 35 years ago | 00:13 |
phogg | maybe somebody decided 32 bits were not enough; machine-id is 128bit. | 00:57 |
Xenguy | All hail the great Posix | 01:00 |
Xenguy | *POSIX | 01:01 |
Xenguy | Did systemd break POSIX? | 01:02 |
Xenguy | ^^ Possibly OT | 01:02 |
fsmithred | looks like it's /usr/bin/hostid which returns a hexadecimal number. The man page just says "print the numeric identifier for the current host" and doesn't say where that number comes from. | 01:09 |
gnarface | no /etc/hostid or /var/adm/hostid anywhere over here. /usr/bin/hostid does return what appears to be a 4-octet hex value though | 02:03 |
rrq | 007F0100 ? | 03:23 |
fsmithred | whoa | 03:24 |
gnarface | rrq: no collision here | 03:24 |
fsmithred | how'd you get my number? | 03:24 |
gnarface | fsmithred: you really have the same number? | 03:24 |
fsmithred | lemme check | 03:25 |
gnarface | i noticed two of my machines had the same first 2 octets | 03:25 |
rrq | the IP of the hostname as of /etc/hosts | 03:25 |
fsmithred | no, but it's really close | 03:25 |
gnarface | i'm not sure where it's getting them but i do suspect there might be insufficient uniqueness to really be able to trust this value | 03:25 |
fsmithred | 007f0101 | 03:26 |
rrq | 127.0.1.1 | 03:26 |
gnarface | my first two octets aren't the same as yours either | 03:26 |
fsmithred | two computers have that number and one has the same as rrq's | 03:26 |
gnarface | a8c02201 on this machine | 03:27 |
gnarface | (ceres) | 03:27 |
rrq | for me it's the ip of the entry in /etc/hosts that is named by hostname | 03:27 |
fsmithred | ascii and jessie here | 03:27 |
rrq | as little-endian integer | 03:27 |
gnarface | a8c02601 on a rpi running ascii .. | 03:28 |
rrq | 192.168.1.38 | 03:28 |
gnarface | e36c2d34 on a laptop running jessie ... | 03:28 |
gnarface | you're saying these are just the ip address of eth0 in hex? | 03:29 |
rrq | yes | 03:29 |
gnarface | hmm. interesting | 03:29 |
rrq | byte swapped | 03:29 |
gnarface | (that's my l4d1&2 server btw) | 03:30 |
gnarface | (have fun) | 03:30 |
fsmithred | my two that are the same currently have addresses in 10.0.0.x and the different one is 192.168.1.x | 03:31 |
fsmithred | but they all started out in 192.168... | 03:31 |
gnarface | (oh there's a chivalry server on there too, i forgot) | 03:31 |
rrq | maybe its what "host $(hostname)" resolves into | 03:32 |
rrq | resolves in /etc/hosts that is | 03:32 |
DocScrutinizer05 | >> Full documentation at: <https://www.gnu.org/software/coreutils/hostid> << | 08:42 |
DocScrutinizer05 | from manpage | 08:42 |
DocScrutinizer05 | locate hostid doesn't reveal anything in etc or elsewhere, except the executable in bin | 08:44 |
DocScrutinizer05 | rrq: actually that's in line with https://www.gnu.org/software/coreutils/hostid >>On that system, the 32-bit quantity happens to be closely related to the system’s Internet address, but that isn’t always the case. << | 08:46 |
ErRandir | I also have 007f0101 and 127.0.1.1 and /etc/hosts | 09:42 |
DocScrutinizer05 | yeah, 007f0100 seems popular, I got one as well | 09:59 |
DocScrutinizer05 | oh, one off | 09:59 |
Leander | https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/gethostid.c;h=0c416f5b8ac0f72d979d6ea073723c9035a0872d;hb=refs/heads/master | 10:10 |
Leander | this is gethostid's source | 10:10 |
Leander | so basically, it takes your hostname, resolves it into your ip address (and the first one it finds is typically 127.0.0.1), then inverts the upper 16 bits and lower 16 bits | 10:11 |
Leander | and because your IP address is stored in big endian format, you have 1.0.0.127 converted into 0.127.1.0, which is 007f0100 in hexadecimal | 10:12 |
KatolaZ | Leander: that is only if /etc/hostid does not exist | 10:17 |
Leander | KatolaZ: yes, I was looking into why most people get the magical 007f0100 | 10:21 |
KatolaZ | oh sure | 10:21 |
KatolaZ | :) | 10:21 |
ErRandir | So I'm better off putting a random number in /etc/hostid, right? | 10:26 |
KatolaZ | ErRandir: I personally think it's pretty useless anyway | 10:53 |
KatolaZ | the most powerful way to avoi being tracked is to disable cookies and javascript | 10:53 |
KatolaZ | and to use tor | 10:53 |
KatolaZ | using tor with cookies enabled is pretty lousy anyway | 10:54 |
iovec | I'm happy devuan is not succumbing to broken ideas like /etc/machine-id | 11:16 |
KatolaZ | xinomilo: your bug report was stuck, but was eventually processed | 14:57 |
sokan | Don't burn me for this question but how can I get the dev1 installation image from rsync mirrors? | 15:07 |
xinomilo | Katolaz: yes, got the aknowledgement allright , thx. | 15:07 |
KatolaZ | sokan: https://devuan.org/get-devuan | 15:17 |
KatolaZ | sokan: look for "RSYNC mirrors"... | 15:17 |
sokan | I meant after that I 'rsync -av mirror.leaseweb.com::devuan/devuan_ascii /path/to/my/desktop/dir`? | 15:27 |
sokan | I ended up `man rsync`. Got me covered I think :P | 15:29 |
sokan | to all these who use a *wm, do you have your usb-stick/devices auto-mounted or do you manually mount it? | 17:28 |
DPA | The website https://devuan.org is down. | 23:40 |
g4570n | yes, but https://files.devuan.org works | 23:41 |
sixwheeledbeast | http://isup.me/devuan.org | 23:58 |
sixwheeledbeast | yep | 23:58 |
golinux | We know | 23:58 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!