I have 4 V440 servers running Solaris 9. I have their interfaces configured (ce0) and have connectivity to our network. However, after reboot,...the connectivity is lost although the interface shows that its still up after an ifconfig -a.

Only after I reconfigure the interface do I restore network connectivity using ifconfig commands from command line.

Would appreciate any help.

Storageguy

Please post a snap from your /var/adm/messages file....
Maybe auto-negoiation isnt working correctly. You can try to hard-code the speed and duplex settings both on the V440 and your switch..

Check that there is a host entry in /etc/hosts for whatever you have in /etc/hostname.ce0

Also if you have a subnetmask that isnt one of the old standard class A,B or C networks it needs to be in /etc/netmasks

Also ... *just in case* check you tcp/ip startup files are present

/etc/rc2.d/S69inet
/etc/rc2.d/S72inetsvc

Networkfre@k, thanks for your reply.....both server and switch are set to 100 Full Duplex.

Unbeliever, thanks for youre reply. Both hosts and hostname.ce0 are correct. I have the subnet listed in the /etc/netmasks file.

/etc/rc2.d/S69inet and /etc/rc2.d/S72inetsvc are there.

Check that /etc/hostname.ce0 is ce(zero) and not capital O which would cause the system not to find the hostname/ip to set up on ce0.

Check that both startup scripts mentioned are executable by root (and owned by root).

Setting ce interface to specific required settings:
The commands for the ce fiber card (Sun GigaSwift 1.0) are:

ndd -set /dev/ce instance 0 (settings for ce0)
ndd -set /dev/ce adv_1000hdx_cap 0 (no half-duplex)
ndd -set /dev/ce adv_pause_cap 1
ndd -set /dev/ce adv_asmpause_cap 0 (send/receive pause frames)
ndd -set /dev/ce adv_autoneg_cap 0 (turn auto-negotiation off)

These should be in a start up file - change to specific settings required by your site. Always check that the interface is truely up by looking with the ndd command parameter link_status (ifconfig can state UP but that does not mean the connection is up).

After rebooting the server, instead of maually putting in the correct values for the server, enter the ifconfig as.

ifconfig ce0 + netmask +

this will tell ifocnfig to use the values in the configuration files for this interface do not plumb the interface before doing this. If it is not plumbed and you get an error, the the problem is most likely caused by the hostname.ce0 having an O instead of a zero as described by RTM.

If however this does work but incorrectly configures the iterface, most probably it will be the netmask which is incorrect. Solaris default is a class b subnet (ffff0000) if you have a class c it should be ffffff00, etc. If the netmask is incorrect add the appropriate entry and try again with the command given above.

Out of interest has it ever worked correctly after a reboot? Also have you recently changed the switch to which it is connected?

We've had problems before with suns and auto negaotiation with switches ... although we've never had it auto negotiate to not up at all :-) We just hard code the switches to the relevant speed now.

If you can and are allowed posting the output of ifconfig -a before and after running reborg's suggested command would tell us a lot.