> > Summary of results > > > > SKAS + TT ./linux-tt-skas > > hangs > > ./linux-tt-skas mode=tt hangs > > TT only ./linux-tt hangs > > > > > > In all cases, boot hangs on sleeping process gets unexpected signal 10 > Ok, now let's wait a moment. Can you run, on the same machine, a vanilla > kernel (even 2.6.10) and verify on that if TT mode works?
Am on it, will be done by tonight > Nobody has ever seen this kind of crash, so I want to see if it's > userspace > related (some strange GCC/glibc/headers problem) or if it's the kernel > triggering it (either a bug in SSI, or a bug of UML which is only > triggered > in that environment). I say "some machine" because distro differences > could > be part of the picture. > > Btw, is the host being part of a cluster at the testing moment or not? The host is a masternode for the cluster, all other nodes are OFF for now but the openSSI is active > Also, on the community page there's a link about these setups (CILinux > seems > related with OpenSSI, if not the same thing): > > http://usermodelinux.org/modules.php?name=News&file=article&sid=185 Yup, but the link above is for combining uml + openssi so that you can set up a "virtual" ssi cluster (several UMLs, one acting as masternode, the others as members of the cluster) In my case, this is the other way around; the host is a open SSI masternode to which ppl will connect to play around with their own UML(s). So the UML boxes won't even know they are working on a SSI cluster and not a single host. > > Btw, is OpenSSI purpose anyhow different from OpenMosix one or not? Just > as a > curiosity about OpenSSI (I'd like to understand a bit more what it does, > I'm > seeing the "features" page). Their global objective is the same; providing a single system image so that you can work on a node and have processes migrating all over the cluster depending on the load of each nodes. OpenSSI uses the same algorithms than open Mosix for load balancing but the implementation is different (less residual information about processes on their initial starting node as far as I understand). Now OpenSSI goes a step farther in my opinion, providing options for cluster file systems and (that's why I use it) incoming tcp/ip connections load balancing (using the LVS project merged with open mosix algorithms). The result is a cluster balancing both processes and connections (thus the node on which processes start). This can be relevant in case of several users trying to compile a kernel (short gcc processes spawned are not lasting long enough to be migrated on the cluster so, if several users compile their own kernels, they can end up clogging 1 single node if they have not be redirected to balance connections) Let me know if I don't make sense, I'll try to be clearer when I post the results of the vanilla 2.6.10 host kernel test. Alessio ------------------------------------------------------- This SF.Net email is sponsored by the 'Do More With Dual!' webinar happening July 14 at 8am PDT/11am EDT. We invite you to explore the latest in dual core and dual graphics technology at this free one hour event hosted by HP, AMD, and NVIDIA. To register visit http://www.hp.com/go/dualwebinar _______________________________________________ User-mode-linux-devel mailing list User-mode-linux-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
