Odd route appears after reboot, Solaris 10 (x4500 series)

[forrie]

We have a point-to-point fiber that connects two thumpers across site which is used for replication of data. Normally this is just a matter of two interfaces, no route and no problem. However, on this system (Solaris 10), when the system reboots, we end up with a route that completely shuts down other traffic. IE:

10.0.0.0 10.111.233.3

Where on other interfaces we use the same IP space like:

10.1.3.4

(ie: 10/8)

It's easy to manually delete the route, our /etc/defaultrouter is correctly set to the primary interface.

I'm more of a Linux person, but I'm wondering if we need to add a persistent route entry for that specific /24 that is not related to our internal networks?

I also noticed, and this may be a side effect of the above, the interfaces are not automatically coming up at boot (ie: ifconfig <interface> up, resolves it). I think it's due to the mishap route.

Any tips would be appreciated.


Thanks.

 

[Annihilannic]

What do you have in /etc/networks (or is it /etc/inet/networks?) for the network in question? Is it the same on both servers?

Annihilannic
[small]tgmlify - code syntax highlighting for your tek-tips posts[/small]

 

[forrie]

/etc/networks is a symlink to /etc/inet/networks Ours is just a standard, there is nothing referencing any other networks there.

Should there be? I read through the manpage, it wasn't clear.

 

[Annihilannic]

Perhaps I'm thinking of netmasks rather than networks.

I'll have a dig around on my Solaris VM when I get a chance.

Annihilannic
[small]tgmlify - code syntax highlighting for your tek-tips posts[/small]

 

[forrie]

Had another issue similar to this, on the secondary system -- the netmask was changed to a /8 (255.0.0.0) when it should be 255.255.255.0 -- this is probably in response to the first system.

This is really p**sing me off. Our Linux systems just run, they don't misbehave like Solaris. I can't figure out out to statically set these so they won't be mucked with.

Did you find anything?


Thanks!

 

[Annihilannic]

Just had a quick look, /etc/inet/netmasks is the file I was thinking of; what does it contain on your systems?

Annihilannic
[small]tgmlify - code syntax highlighting for your tek-tips posts[/small]

 

[forrie]

/etc/netmasks is symlinked to /etc/inet/netmasks, and we have:

10.250.0.0 255.255.255.0
10.251.0.0 255.255.255.0
10.245.137.0 255.255.255.0

So you can see, when Solaris puts in a 10/8 route for the last network, it shuts everything else down.


Thanks.

 

[forrie]

Again this only happens after the system reboots. The correct route we have for the point to point config is:

Routing Table: IPv4
Destination Gateway Flags Ref Use Interface
-------------------- -------------------- ----- ----- ---------- ---------
default 10.250.0.1 UG 1 4669
10.245.137.0 10.245.137.20 U 1 612 e1000g2

I'm not sure why Solaris won't keep this information --- where can I place this so that it stays static?