On Tue, Feb 24, 2004 at 09:40:27PM -0500, Michael Gargiullo wrote: > I find this thread very interesting. We run several multiproc boxes on > FC1 without a hitch. I will say, I don't think any of them are overly > stressed, I like to over build when I have the chance. > > Tyan MB with AMD chips, and Dell 2650's dual Intel Xeon w > hyperthreading. I have on of those, amongst others, and I can get it to lockup at will. > We run the FC1 SMP kernels... Should I be looking for anything specific > to go wrong? These are the bugzillas: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=109962 https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=109497 The thing, in fact the only thing I can see, that triggers this bug is a umount. http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=109962#c40 has a script that continuously mount/umount's a small loop-back file system. Hangs my FC1 SMP systems every time. I have got also got hangs when a new kernel is installed. One of the last steps of this is the creation of the initrd. That involves a loop back mount, and an umount. The kernel sometimes locks up on the umount. My impression is that the lockup is more likely to occur if there is some disk activity at the time. -- Norman Gaywood, Systems Administrator School of Mathematics, Statistics and Computer Science University of New England, Armidale, NSW 2351, Australia norm@xxxxxxxxxxxxxxxxx Phone: +61 (0)2 6773 2412 http://turing.une.edu.au/~norm Fax: +61 (0)2 6773 3312 Please avoid sending me Word or PowerPoint attachments. See http://www.fsf.org/philosophy/no-word-attachments.html
