An intermediate summary;
I increased the max threads in kernel ( /etc/system:
nfs:nfs3_max_threads=64 ) and in /etc/init.d/nfs.server: nfsd -a 256
(Thomas & Greg & Casper ) Now the server is more efficient ;-)
Casper, recommended me to use truss/pstack on this problem. Tracing from the
beginning, I discovered that when automount unmounts the unused home
directory for predefined 10 minutes of time, it can never mount it again.
truss/pstack showed that it does not send any mount requests to mountd on
remote server. On server, the mountd, does not receive anything. nis+ works
fine, no problem...
After login process completes successfully, it tries to "cd /home/userdir".
Automountd gets in deadlock. If I restart automountd, it recovers. After
logout, and unmount after 10 minutes, user can not log in to that machine,
In normal operation automountd initiates the mount, that is, mount proc
checks resolv.conf, hosts, does nis+ door queries, and attempts the mount.
An awful but working temporary fix is putting a automountd restart code in
crontab (<10min). :-(
I checked if there are any patches available for automountd for sol 2.7, I
found that 107127-02 patch is already installed. Although this patch only
fixes : 4188020 (Mar-12-99): mount on /etc/dt is cached and will not be
replicated, that it does not fix what I had wished it to.
-- Emre Celebi Bogazici University, Department of Computer Engineering CNRC Computer Networks Research Center Phone:+90 212 2631540 ext 2125 "Perilous to all of us are the devices of an art deeper than we ourselves possess." - Gandalf The Grey, J.R.R. Tolkien, "Lord of the Rings"
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:14:07 CDT