Home » General » Support » *SOLVED* /dev destroyed upon gnome install
| Re: /dev destroyed upon gnome install [message #10832 is a reply to message #10797] |
Mon, 05 March 2007 14:34   |
Vasily Tarasov
Messages: 1345
Registered: January 2006
|
Senior Member |
|
|
Hello,
| Quote: |
But is it correct to have private/ID/dev normally poulated?
|
Sorry, I do not clearly understand your question.  There are several possibilities to populate /dev:
1) static content
2) udev
3) devfs, but it is obsolete, so we don't take this possibility into account
All these ways are "normal", and you'll get "normal" content. In some situations one way is preferable, but in other - other one. In OpenVZ, static content is a simple and logical choice, because sysfs is not completely virtualized, and therefore udev can't be used simply. But as you can see, many packets need udev by dependency. In RPM-based distribution such dependency lack can be stoppled with vzdev package. But in Debian we can't do so... (or can we? Actually I'm not very familiar with Debian, but AFAIK it isn't possible to do it easy.) So the simple way in Debian is to install udev to satisfy all dependencies, and after that work around /dev content to get VE work in the way we wish! 
| Quote: |
I am afraid but I can not quickly give access to the HW node.
You mean by ssh?
I packed the vps but my public space it's not wide enough can I ftp somewhere?
About the:
# vzctl set <veid> --features sysfs:on --save
what you mean by "does it work now?". I still can't ssh to it.
I can vzctl enter it. But it was my fault previously reporting this as not working, I forget to report this in previous post, apologies.
|
Ooops! So you can enter it?! It changes the situation dramatically!.. So you have the same situation as I have! And I was digging into the direction why I can enter VE and you can't do it! Next time, please, be a bit more precise...
Ok then, what problems do you have with Debian VE now? You can't ssh in it? If so, then I don't need an access to your node and I don't need your gziped VE, because I know the reason.
After installing gnome in Debian VE and restarting this VE, I enter it via `vzctl enter` command and do the following:
debian-ve# ps aux | grep sshd
debian-ve# /etc/init.d/ssh start
Starting OpenBSD Secure Shell server: sshdPRNG is not seeded
It makes me think, that the problem is in /dev/random or in /dev/urandom. Checking this guess:
debian-ve# mknod /dev/random c 1 8
debian-ve# /etc/init.d/ssh start
Starting OpenBSD Secure Shell server: sshd.
debian-ve# echo $?
0
debian-ve# ps aux | grep sshd
So sshd still can't start (but because of some other problem). In logs I find:
sshd[1688]: fatal: daemon() failed: Success
So, I look at man page on daemon():
The glibc implementation can also return -1 when /dev/null exists but is not a character device with the expected major and minor numbers. In this case errno need not be set.
This is it! Let's create /dev/null:
debian-ve# mknod /dev/null c 1 3
debian-ve# /etc/init.d/ssh start
Starting OpenBSD Secure Shell server: sshd.
debian-ve# ps aux | grep ssh
root 1819 0.0 0.0 3468 920 ? Ss 14:16 0:00 /usr/sbin/sshd
Now I can ssh in VE!
The next problem is that after restarting VE /dev/random and /dev/null dissappear (thank to udev). How to fight against it in my next post. This one is too big.
[Updated on: Mon, 05 March 2007 14:36] Report message to a moderator |
|
|
|
Current Time: Mon Jul 21 12:23:04 GMT 2025
Total time taken to generate the page: 0.07929 seconds
|
OpenVZ Forum
Members
Search
Help
Register
Login
Home
