|00:34||roca has left IRC (firstname.lastname@example.org, Quit: roca)|
|00:44||vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc)|
|01:06||lucascastro has joined IRC (email@example.com)|
|01:16||lucascastro has left IRC (firstname.lastname@example.org, Remote host closed the connection)|
|01:16||lucascastro has joined IRC (email@example.com)|
|02:26||vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Quit: leaving)|
|02:43||ogra_ has left IRC (firstname.lastname@example.org, Ping timeout: 260 seconds)|
|02:44||ogra_ has joined IRC (email@example.com)|
|05:18||kidar has left IRC (firstname.lastname@example.org, Quit: Leaving)|
|05:21||kidar_ has joined IRC (email@example.com)|
|05:42||lucascastro has left IRC (firstname.lastname@example.org, Remote host closed the connection)|
|05:42||lucascastro has joined IRC (email@example.com)|
|06:19||jgee has left IRC (firstname.lastname@example.org, Ping timeout: 252 seconds)|
|06:28||Statler|Home has joined IRC (Statler|Home!~Georg@p5B30EA7C.dip0.t-ipconnect.de)|
|06:51||ricotz has joined IRC (ricotz!~ricotz@p5B2A9D27.dip0.t-ipconnect.de)|
|06:51||ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz)|
alkisg: this morning, my fat client was frozen, could still switch consoles, but everything that seemed to require file system access blocked indefinitely. I suspect something with NBD, but it's not clear of course. Is there anything meaningful for me to do in order to debug this if it happens again?
fiesh: nfs is more tolerant in client disconnections and reconnections. NBD isn't; it requires a persistent connection. Is it possible that the client was disconnected for a while?
In general, it's possible to leave a client on for any number of days, but you can't e.g. pull the network cable for an hour and reconnect it...
Check the nbd-server processes at that point, if the client still has a connection or not:
client-list: to get a list of all nbd-clients (which sometimes is the same as ltsp clients), run: ss -n | sed -n 's/.*:10809 *\([0-9.]*\):.*/\1/p' | sort -Vu
alkisg: no, it's not been disconnected
fiesh: and I would assume no standby time either... do you have a client shell in some vt? Can you run dmesg on the client?
but I'll check that list the next time it happens
alkisg: I couldn't even log in any more, so I'll open up a shell now so that I have something to go to -- however, I guess it won't be possible to run dmesg
ahh there's another one!
and this one gives an error
packet_write_wait: Connection to <ip address> 22: Broken pipe
so two clients (all that were booted with the fat client so far) stopped working over night
but nothing else that can be done with it, no log in, no dmesg, nothing :(
fiesh: port 22 is ssh, not nbd
NBD is at port 10809
Ah maybe that's the error message and not the port...
If we could look at `dmesg`, and we saw "link up/link down" messages, then we'd know if it was a disconnection issue...
fiesh: it would also be helpful to have a virtual (vbox) client up and running for a few days. The vbox client is guaranteed not to have connection issues. If the others fail and the vbox one doesn't fail, that would point us to connection issues.
hmmm easier said than done, the server doesn't have any graphical user interface installed
but you are right, that's an ssh message
it is the port
so the server's dmesg is flooded with [5268299.855930] NFSD: client 192.168.3.117 testing state ID with incorrect client ID
which unfortunately seems to have overwritten any other messages that might have occured
fiesh: you can boot a thin client, arrange a vbox thin client instead the real thin client, and finally log in to the server locally and launch it headless via vboxmanage
I can help with the thin client test if you want
Re: NFSD, I'm not using NFS for /home... I should... but it shouldn't be any different from your previous setup in that part
I.e. I don't have much NFS experience to relay here
this is about the error message
oh yeah, I can just use a thin client to begin with, you're right
fiesh: if you think that the clients changing IPs and hostnames may cause NFS issues, it's possible to define static hostnames in many places
One place is lts.conf. Another is to tell the client to use a hostname based on MAC, not IP.
And of course the DHCP server can specify the hostname
hmm I don't think they're changing IP addresses, I see no reason for that, but I'll try to keep an eye open
Btw about the ssh message... if ssh complains about a broken pipe, that's 99% usually true, i.e. there was indeed a form of disconnection
hmm maybe our switch went to funky town
oh but the NFS messages are all from yesterday morning
so they cannot be related to the disconenct
And ssh is a lot more tolerant than nbd
and after them, dmesg logs no device down
What's the topology? server <=> single switch <=> clients?
and the switch is on the same UPS as the server
so it cannot have had a power down
hmm ok, I'll just see if the problem persists or not
if it's the same every morning, I'll have to look into it or tell everybody to log out over night ;)
Ah, here's an idea. Do you have this file on the server? /usr/share/ldm/rc.d/I01-nbd-checkupdate
no, only two X... files
OK, good, then that's not to blame
|08:01||wim1 has joined IRC (wim1!~Thunderbi@WEGC203033.UNI-GRAZ.AT)|
|08:15||ZAJDAN has joined IRC (ZAJDAN!4d30954b@gateway/web/freenode/ip.220.127.116.11)|
|08:52||Statler|Home has left IRC (Statler|Home!~Georg@p5B30EA7C.dip0.t-ipconnect.de, Remote host closed the connection)|
|09:31||Statler_Office has joined IRC (Statler_Office!~Georg@gwrz3.lohn24.de)|
|10:26||root^PC has joined IRC (root^PC!d9d98abb@gateway/web/freenode/ip.18.104.22.168)|
|10:26||root^PC has left IRC (root^PC!d9d98abb@gateway/web/freenode/ip.22.214.171.124)|
|11:40||ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Quit: Leaving)|
|11:42||ricotz has joined IRC (ricotz!~ricotz@ubuntu/member/ricotz)|
|11:53||Faith has joined IRC (Faith!~Paty_@unaffiliated/faith)|
|11:54||wim1 has left IRC (wim1!~Thunderbi@WEGC203033.UNI-GRAZ.AT, Quit: wim1)|
|11:56||wim1 has joined IRC (wim1!~Thunderbi@WEGC203033.UNI-GRAZ.AT)|
|12:54||wim1 has left IRC (wim1!~Thunderbi@WEGC203033.UNI-GRAZ.AT, Quit: wim1)|
|13:38||kidar_ has left IRC (email@example.com, Quit: Leaving)|
|13:42||ben_roose has joined IRC (firstname.lastname@example.org)|
|13:49||nehemiah has joined IRC (email@example.com)|
I was just wondering if it would theoretically be possible to write the contents of the LTSP image to a USB thumb drive or a internal hard drive, install grub to it and make it boot able and working as a standalone installation?
local-boot: If you want LTSP fat clients on a low-speed network, you can put i386.img on e.g. C:\Boot\LTSP\i386.img and use this command line in pxelinux.cfg: APPEND ro initrd=ltsp/i386/initrd.img init=/sbin/init-ltsp root=/dev/sda1 rootflags=ro loop=/Boot/LTSP/i386.img; IPAPPEND 3
But would that be able to work w/o the ltsp server?
probably not. it's not hard to make an ubuntu bootable image with whatever you want on it
No that's true, it's just an idea I had.
nehemiah: yes of course, that's why I wrote that note there
alkisg: if it can't reach the server, wouldn't login fail?
Ah he said w/o, those damn shortcuts
Well if it's an ltsp-pnp image, it should work fine, as long as you dont pass init=/sbin/init-ltsp, when you don't have the server nearby
I.e. it will work in the same exact sense that the server itself works
With LightDM or GDM for the login manager, with local users etc etc
Sorry for that. Yeah I kind of liked the idea of havinf somebody boot is laptop of the LTSP server and then be able to run a script that would set that laptop up with more or less the identical set-up.
But I can create a separate image for that.
nehemiah: if you're using ltsp-pnp or ltsp-manager, you don't need a separate image
I'm using LTSP manager
The clients boot from the server image/template anyway; they have ltsp-server and dnsmasq and everything installed, just not activated
OK, so yeah what you asked is completely possible
You can even write it to a real partition, and it would be the same as if you had cloned the server, except for the --cleanup phase, if you're using that
E.g. I could write a script where you would select a client in epoptes, go to the Execute command menu, and it would install the ltsp client chroot as a local installation there
All with one command via epoptes
Wow, thank you for that. that sounds brilliant.
I'll look into that
|14:13||Statler_Office has left IRC (Statler_Office!~Georg@gwrz3.lohn24.de, Remote host closed the connection)|
|14:14||Statler_Office has joined IRC (Statler_Office!~Georg@gwrz3.lohn24.de)|
|14:18||Statler_Office has joined IRC (Statler_Office!~Georg@gwrz3.lohn24.de)|
|14:54||nehemiah has left IRC (firstname.lastname@example.org, Remote host closed the connection)|
|15:03||JerryT has left IRC (JerryTemail@example.com, Quit: Leaving)|
|15:18||jgee has joined IRC (firstname.lastname@example.org)|
|16:02||roca has joined IRC (email@example.com)|
|16:03||mondeydiver has left IRC (firstname.lastname@example.org, Read error: Connection reset by peer)|
|16:08||roca has left IRC (email@example.com, Quit: roca)|
alkisg: ok, I have two more minor issues that maybe you can help me with
a) /etc/cups/client.conf vanishes -- it's present in the squashfs, but after booting, it's gone, meaning printing doesn't' work... I have no idea how that can be, something must delete it in the overlay? Can I somehow track that?
b) before ldm starts, it says (took me a while to catch this with ctrl+s at the right millisecond):
File "/usr/share/ldm/ldminfod", line 186
print "ERROR: failed to run locale"
SyntaxError: Missing parantheses in call to 'print'
maybe this is a python 2 vs 3 issue?
(locale is present and this appears to be a purely syntactical problem)
fiesh: CUPS_SERVER=$SERVER which is the default, or localhost, make the ltsp init scripts overwrite cups.conf
About ldminfod, let me check locally...
fiesh: hmm, I don't get any errors running ldminfod in my vbox gentoo...
If you run it manually on the client, do you see that error?
Also, try running just `locale`, and see if it outputs a LANG= line
alkisg: ah, I didn't know about the CUPS lts option, I'll give it a shot
fiesh: I don't see any option NOT to generate cups/client.conf; do you need something other than the server in that file?
alkisg: and yes, try python2 vs python3 with ldminfod
alkisg: it works with python2 but throws this error with 3
I think it should probably be fixed, I guess print "bla" needs to be chagned to print("bla"), without being a python programmer
|16:48||ZAJDAN has left IRC (ZAJDAN!4d30954b@gateway/web/freenode/ip.126.96.36.199, Ping timeout: 260 seconds)|
fiesh: you have python symlinked to python3? Isn't that bad?
yes and no
When a program has #!/usr/bin/python, isn't it supposed to run with python2, while if it has python3, with python 3?
I don't think so, gentoo allows you to select the system python version and symlinks python to it
so far I have not had problems
(and I've had it for quite a while)
OK... since ldminfod hasn't been ported to python 3, the first line would need to read python2 in your system
for the time being, all distributions should ensure that python refers to the same target as python2.
|16:55||Statler_Office has left IRC (Statler_Office!~Georg@gwrz3.lohn24.de, Remote host closed the connection)|
I think that's the preferred way
py3 has been out for 10 years now
I think in general it makes sense to keep python a pointer to the older release, even when python 4 and python 5 arrive in the future...
so each .py file should specify a number in which it runs with
|17:09||lucascastro has left IRC (firstname.lastname@example.org, Remote host closed the connection)|
ah I see, then gentoo doesn't follow this recommendation I suppose
making it compatible with both 2 and 3 is doable btw, https://pypi.python.org/pypi/six works like a charm
Just making ltsp python3-only, would be fine, if someone had the time for it :)
Porting small program like ldminfod from py2 to py3 is quite easy even without six. In py2 you can use __future__ imports for print() and similar.
It could run on both py2 and py3 easily too.
|18:25||lucascastro has joined IRC (email@example.com)|
|19:01||roca has joined IRC (firstname.lastname@example.org)|
|19:47||roca has left IRC (email@example.com, Ping timeout: 246 seconds)|
|20:16||ben_roose has left IRC (firstname.lastname@example.org, Remote host closed the connection)|
|20:28||vagrantc has joined IRC (vagrantc!~vagrant@unaffiliated/vagrantc)|
|20:34||Faith has left IRC (Faith!~Paty_@unaffiliated/faith, Quit: Leaving)|
|20:47||lucascastro has left IRC (email@example.com, Remote host closed the connection)|
|22:19||roca has joined IRC (firstname.lastname@example.org)|
|22:54||vagrantc has left IRC (vagrantc!~vagrant@unaffiliated/vagrantc, Quit: leaving)|
|22:58||ricotz has left IRC (ricotz!~ricotz@ubuntu/member/ricotz, Quit: Leaving)|
|23:41||roca has left IRC (email@example.com, Quit: roca)|