Jeff,

Yeah, I have included these ports for the rules set for 
connectivity to our NetBackup server.  

I also allow access to the range of ports from 512 
to 1024 as the range of "random" ports required
by netbackup:

Veritas NetBackup Server - juniper0.ssd.census.gov 
# -  4 ports from /etc/services
pass in log quick on eri0 proto tcp from x.x.x.x to loghost port = 13720 fl
ags S keep state
pass in log quick on eri0 proto tcp from x.x.x.x to loghost port = 13722 fl
ags S keep state
pass in log quick on eri0 proto tcp from x.x.x.x to loghost port = 13782 fl
ags S keep state
pass in log quick on eri0 proto tcp from x.x.x.x to loghost port = 13783 fl
ags S keep state 
# 2003 January 23 - jthomas
# Added -- Having socket failures with backup.
pass in log quick on eri0 proto tcp from x.x.x.x to loghost port = 13724 fl
ags S keep state 
# - range of ports selected randomly by NetBackup server
pass in log quick on eri0 proto tcp from x.x.x.x to loghost port 512 >< 1024
flags S keep state


The problem is that when the box has ipfilter enabled with this
configuration,the NetBackup Administrators report that the
full backups are failing due to socket timeout errors.  I'm 
resolved to believe that it's a problem with IPFilter because
full backups are able to complete when packet filtering is
turned off.

Since we already have firewalling on our networks, I could just 
open up all ports to the Netbackup server, but I really would
like to get this thing right.  Any more suggestions?

John

Reply via email to