Hi,
I am trying to add an exec host (dancer) to a working cluster, but
master (fedora1) seems to have problems resolving host name. In the logs:
fedora1|E|commlib error: can't resolve host name (can't resolve rdata hostname
"dancer.uscf.edu")
fedora1|E|commlib error: local host name error (remote rdata host name "dancer.uscf.edu"
is not equal to local resolved host name "(HOST_NOT_RESOLVABLE)")
qping from new exec host to qmaster:
$ qping fedora1 6444 qmaster 1
endpoint fedora1/qmaster/1 at port 6444: can't find connection
access denied: server host resolves rdata host "dancer.uscf.edu" as
"(HOST_NOT_RESOLVABLE)"
endpoint fedora1/qmaster/1 at port 6444: can't find connection
whereas if I start execd on dancer, qping from qmaster to dancer's execd
works fine.
/etc/hosts on both machines has:
xx.xx.xx.yy fedora1 fedora1.ucsf.edu
xx.xx.xx.xx dancer dancer.ucsf.edu
I tried to create host_aliases with an entry
dancer dancer.ucsf.edu xx.xx.xx.xx
but didn't help.
/etc/resolv.conf has
domain ucsf.edu
search ucsf.edu
on both machines.
$SGE_ROOT/utilbin/linux-x64/gethostby{name,address} report correct names
and ip from both siders.
I tried to turn the firewalls off, but it makes no difference.
Appreciate any constructive advice!!
Miro
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users
