Netmask is set to 255.255.0.0 while it should be set to 255.255.255.128
Thomas Glanzmann
I have a Solaris 10 Update 7 machine which sets the netmask and network
route incorrectly. And I have two questions that I can't answer myself:
1) As far as I can tell /etc/netmasks, /etc/hosts and /etc/hostname.* are
configured correctly but the netmask is set wrong
deerlf0x68-root # ifconfig nxge0
nxge0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
inet 157.163.247.48 netmask ffff0000 broadcast 157.163.255.255
groupname mpathd1
ether 0:21:28:42:b9:7a
deerlf0x68-root # cat /etc/hostname.nxge0
deerlf0x68 group mpathd1 up
deerlf0x68-root # cat /etc/hostname.nxge4
group mpathd1 standby up
deerlf0x68-root # grep -v '^#' /etc/netmasks
157.163.247.0 255.255.255.128
deerlf0x68-root # netstat -rn
Routing Table: IPv4
Destination Gateway Flags Ref Use Interface
-------------------- -------------------- ----- ----- ---------- ---------
default 157.163.247.126 UG 1 34
157.163.0.0 157.163.247.48 U 1 78 nxge0
192.168.224.1 192.168.224.2 UH 1 1 sppp0
224.0.0.0 157.163.247.48 U 1 0 nxge0
127.0.0.1 127.0.0.1 UH 3 129 lo0
deerlf0x68-root # cat /etc/defaultrouter
157.163.247.126
2) Even if the netmask is configured wrong the machine seems to use the
defaultrouter for everything outside /25 for some reason. It can reach for
example my workstation which is not on the same network.
deerlf0x68-root # traceroute -n ad054764pc
traceroute: Warning: Multiple interfaces found; using 157.163.247.48 @ nxge0
traceroute to ad054764pc (157.163.208.118), 30 hops max, 40 byte packets
1 157.163.247.124 0.330 ms 0.399 ms 0.203 ms
2 10.112.11.36 1.060 ms 0.858 ms 0.856 ms
3 10.112.11.226 1.059 ms 1.075 ms 1.076 ms
4 10.112.11.17 1.714 ms 1.512 ms 1.511 ms
5 172.16.65.2 1.714 ms 1.515 ms 1.510 ms
6 157.163.208.118 1.493 ms 1.514 ms 1.511 ms
The default router answers as 157.163.247.124 because we have HSRP (Hot Standby
Router Protocol (Cisco, IOS)) on that network enabled.
Thomas
John D Groenveld
Thomas Glanzmann <tho...@glanzmann.de> wrote:
>deerlf0x68-root # grep -v '^#' /etc/netmasks
>157.163.247.0 255.255.255.128
Solaris may expect you to be pedantic in netmasks(4):
157.163.0.0 255.255.0.0
157.163.247.0 255.255.255.128
157.163.247.128 255.255.255.128
John
groe...@acm.org
Thomas Glanzmann
* John D Groenveld <groe...@cse.psu.edu>:
> Thomas Glanzmann <tho...@glanzmann.de> wrote:
>>deerlf0x68-root # grep -v '^#' /etc/netmasks
>>157.163.247.0 255.255.255.128
> Solaris may expect you to be pedantic in netmasks(4):
> 157.163.0.0 255.255.0.0
> 157.163.247.0 255.255.255.128
> 157.163.247.128 255.255.255.128
thank you, I'll modify the netmasks and try again. The problem is that
the system is now in production and I may have to wait a bit until I can
reboot it. However what I don't get is that I sys-unconfigured a Solaris
10 Update 8 x86 and than configured the network using the wizard. And
there it did _not_ put the three entries in, but it worked. Than I
modified hostname.e1000g{0,1} to do active/passive with link status and
the netmask was still right. Might be possible that I miss something
else?
Thomas
Thomas Glanzmann
* Thomas Glanzmann <tho...@glanzmann.de>:
> thank you, I'll modify the netmasks and try again. The problem is that
> the system is now in production and I may have to wait a bit until I can
> reboot it. However what I don't get is that I sys-unconfigured a Solaris
> 10 Update 8 x86 and than configured the network using the wizard. And
> there it did _not_ put the three entries in, but it worked. Than I
> modified hostname.e1000g{0,1} to do active/passive with link status and
> the netmask was still right. Might be possible that I miss something
> else?
here is a screenshot (sorry that was a minimal installation without ssh)
from the machine were I did try to reproduce the problem:
http://thomas.glanzmann.de/tmp/screenshot-mini-2009-12-11-19:08:03.png
As you can see the subnetting is fine and there is only one entry in
netmasks. Strange
Thomas
Diego
> the system is now in production and I may have to wait a bit until I can
> reboot it.
However, it's not necessary to reboot the system, just open the
console, and make a:
ifconfig name_interface hostname_or_ip/your_subnet_decimal_notation
broadcast + up
And it will go up with the correct settings. Just expect to have a
little instability until the new ip/netmask will come up.
And if you want to just test the correct sintax in your /etc/netmasks,
do:
ifconfig name_interface hostname_or_ip netmask + broadcast + up
And it should just take the correct value from the file.
Have a nice day.
Diego.
Darren Dunham
> deerlf0x68-root # cat /etc/hostname.nxge0
> deerlf0x68 group mpathd1 up
That's odd. Why do you have 'up' in there?
Can you show the output of 'wc /etc/hostname*'?
--
Darren
Thomas Glanzmann
* John D Groenveld <groe...@cse.psu.edu>:
> Solaris may expect you to be pedantic in netmasks(4):
> 157.163.0.0 255.255.0.0
> 157.163.247.0 255.255.255.128
> 157.163.247.128 255.255.255.128
even if I could not reproduce it on another machine, I added the entries
and today got the timeslot to reboot the machine and it works:
deerlf0x68-root # tail -3 /etc/netmasks
157.163.0.0 255.255.0.0
157.163.247.0 255.255.255.128
157.163.247.128 255.255.255.128
deerlf0x68-root # ifconfig nxge0
nxge0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
inet 157.163.247.48 netmask ffffff80 broadcast 157.163.247.127
groupname mpathd1
ether 0:21:28:42:b9:7a
So to formulate as general rule: If you use CLASS A, B or C networks you can
just add the netmask in questions, if you use another netmask you have to
specify the hirachy up to the next CLASS A, B or C network, is that correct?
Thomas
Darren Dunham
>
> So to formulate as general rule: If you use CLASS A, B or C networks you can
> just add the netmask in questions, if you use another netmask you have to
> specify the hirachy up to the next CLASS A, B or C network, is that correct?
Don't think so.
You will also see the problem you describe if your hostname.nxge0 file
had a blank line in it.
--
Darren
Thomas Glanzmann
* Thomas Glanzmann <tho...@glanzmann.de>:
> 1) As far as I can tell /etc/netmasks, /etc/hosts and /etc/hostname.* are
> configured correctly but the netmask is set wrong
> Thomas
ulm:/root/root 1 : getent netmasks 157.163.247.0
157.163.247.0 255.255.0.0
ulm:/root/root 2 : grep netmasks /etc/nsswitch.conf
netmasks: nis [NOTFOUND=return] files
netmasks was configured to retrieve them using NIS and NIS was misconfigured.
Thomas