From Bugzilla Helper: User-Agent: Mozilla/5.0 (compatible; Konqueror/3.2; Linux) (KHTML, like Gecko) Description of problem: nfs mounts from this kernel's nfsd just hang forever. cat /etc/exports / 139.85.0.0/16(rw,async) *.hns.com(rw,async) 10.100.20.0/16(rw,async) *.adg.labs.md.hnsnet(rw,async) 10.100.21.0/16(rw,async) No error messages, just hang. In fact, no message at all, not even normal "mount request...". Here is an example, I'm telnetting to a client which should automount my home (I've been using this same setup forever, there is nothing new except the nfs server kernel): Red Hat Enterprise Linux WS release 3 (Taroon Update 1) Kernel 2.4.21-9.0.1.ELsmp on an i686 login: nbecker Password: [... still hung forever..., no message at all on server log...] Version-Release number of selected component (if applicable): 2.6.4-1.281 How reproducible: Always Steps to Reproduce: 1. attempt to mount 2. 3. Additional info:
Workaround: If I remove all overlapping entries from /etc/exports the problem is gone. That is, in my previous exports, I had both IP address ranges and names that mapped to the same IPs. I got rid of the names and kept only IP ranges, and now it's OK. IMHO, this fix is totally unacceptable. It should work with this workaround.
Talk to the nfs-utils maintainer upstream. Parsing of the exports file is kinda dicey.
would it be possible to post a AltSysRq-t backtrace of all the the processes which would help tell me where the nfsds are hanging.
this still a problem with the 2.6.9 kernel ?
No problem now.