Adam Carheden wrote:
>I''m using Xen as a development enviornment for Gentoo AMD64
cluster. I
>have a master node and two slave nodes, all DOMU. All are using the
>same kernel, but when I try to log on to the slave nodes via SSH, the
>client hangs and the server reports
>
>Jan 4 17:14:17 64node02-test sshd[10238]: Accepted publickey for root
>from 10.1.10.206 port 48333 ssh2
>Jan 4 17:14:19 64node02-test sshd[10238]: error: openpty: No such
>file or directory
>Jan 4 17:14:19 64node02-test sshd[10238]: error: session_pty_req:
>session 0 alloc failed
>Jan 4 17:14:19 64node02-test sshd(pam_unix)[11264]: session opened
>for user root by (uid=0)
>
>I can log on successfully via a xen console or using ssh
>`64node01-test bash -i`, so the problem is with the slave nodes
>allocating /dev/pts/X.
>
>- The actual cluster (non-xen) works fine.
>- I''ve tried openssh 3.9_p1-r2 and 4.2_p1 (both Gentoo) with the
same results
>- The slaves and master are using the same xen kernel.
>- The problems happens both with my xenU kernel and my xen0 kernel.
>- The master and slaves are (in theory) identical except that the
>slaves boot from an nfs share on the master (i.e. I use ''root
>"root=/dev/nfs ro"'' as opposed to disk = [ nfs:...] in my
xen config
>file)
>- Both the master and slaves report devpts in mount:
> devpts on /dev/pts type devpts (rw)
>
>Any one have any hints for me?
>
>Thanks
>--
>Adam Carheden
>
>_______________________________________________
>Xen-users mailing list
>Xen-users@lists.xensource.com
>http://lists.xensource.com/xen-users
>
>
>
>
I was going to suggest that your kernel is not set up right for pts, but
then I read you do have devpts mount properly. Here is my kernel config
anyway for Filesystems->Pseudo Filesystems:
[ ] /dev file system support (OBSOLETE)
[*] /dev/pts Extended Attributes
[*] /dev/pts Security Labels
Gentoo switched from devfs to udev not too long ago and it tripped up a
lot of people, myself included.
As a datapoint for you, my gentoo AMD64 xen installation does not have
these pty problems.
Regards,
Dan.
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users