I am doing some tests on a kickstart install of RHEL4 and discovered a problem with my NFS server, after a large amount of data being transferred it will lock up the ethernet interface and refuse to talk to the NFS client until the interface has been deconfigured with "ifconfig down" and then configured again.
When this problem occurs it stops the install (as expected), but it also stops the mouse cursor from moving. This seems to be a bug to me, the X server should not be accessing files on the NFS server (as far as I am aware) and therefore there is no good cause for the mouse cursor to stop.
Once the NFS server starts responding again the install proceeds and the mouse cursor responds.
Is this considered to be working as designed or is it a bug?
Russell Coker wrote:
I am doing some tests on a kickstart install of RHEL4 and discovered a problem with my NFS server, after a large amount of data being transferred it will lock up the ethernet interface and refuse to talk to the NFS client until the interface has been deconfigured with "ifconfig down" and then configured again.
Have you investigated plain old hardware problems on the nfs client and server. You should never have to re-config the interface to get it to start working again.
Cheers,
On Sunday 27 February 2005 06:12, Eric Warnke eric@snowmoon.com wrote:
Russell Coker wrote:
I am doing some tests on a kickstart install of RHEL4 and discovered a problem with my NFS server, after a large amount of data being transferred it will lock up the ethernet interface and refuse to talk to the NFS client until the interface has been deconfigured with "ifconfig down" and then configured again.
Have you investigated plain old hardware problems on the nfs client and server. You should never have to re-config the interface to get it to start working again.
True, there are obviously some issues with the server, and I will be investigating them ASAP. But in the mean time the client doesn't seem to be operating in the way I expect.
Alexandre has posted a message that might explain the problem, although I'm surprised that a machine with 256M of RAM would want to discard executable pages from the X binary on an install.
On Feb 26, 2005, Russell Coker russell@coker.com.au wrote:
When this problem occurs it stops the install (as expected), but it also stops the mouse cursor from moving. This seems to be a bug to me, the X server should not be accessing files on the NFS server (as far as I am aware) and therefore there is no good cause for the mouse cursor to stop.
If the X server code is mounted from the NFS server and it needs swapping in, then the only place you can get it from is the NFS server.
devel@lists.stg.fedoraproject.org