>Reading this message and your original report, I can't seem to determine >under exactly what conditions this problem is supposed to occur. Does the >problem occur with > >- amd64 clients talking to amd64 servers? >- amd64 clients talking to i386 servers? >- i386 clients talking to amd64 servers? >- Some other combination? > > >In each case, exactly what OpenAFS versions are the client and server? >If the fileserver crashes, does it leave a core file behind? >If so, run gdb on that fileserver binary and core file, type 'where', and >send the results. I have seen the problem while trying to setup a new server (new cell) on AMD64. When trying to connect to this server I see the problem, no matter if the client is running on the same node or on a different one, but all AMD64, all running the same version of OpenAFS. However, I have no problems connecting the client to an _existing_ cell. For the latest test I used the released OpenAFS 1.3.80 tar ball from the OpenAFS side. I made a complete protocol of the last attempt, which I can send you personally when I'm back in my office on monday (don't know if this is helpful). The fs command does not core-dump, but simply gets stuck, so nothing I could use to run gdb on it.
_______________________________________________ OpenAFS-devel mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-devel
