UNIX for Dummies Questions & Answers

Whenever i start my server it takes app 2 minutes to get into network (After starting the network service on boot).

I inquire the network guys for possible causes but they say it will need both side monitoring .
One is switch side and the second is server side.

I m not sure how do i check network logs (do i need to enable them at kernel level?(noob here))

I have bonded the network interfaces with LACP protocol.

what do you mean by:

OOh i made it little confusing .

Actually what i mean to say is this.
"I start the Server .It loads the kernel and then start running services by its own"
whenever it starts the service of network for initializing the network interface and bonding it says the status as ok.
But when i try to ping the server from different machine it remains unpingable .
After few minutes it comes to the network.

Hope i made my point clear.

That may be loopback interface...
then it has to do some cleanup...then initialize all the loggers, then configure Ethernet interfaces, followed by lan interfaces and only after all that will start all internet daemons...
Only now will ping work...
Then it will go multi-user and people can start connecting...
2 minutes is not short... but its not long either... To shorten the time you could do some cleaning up before the reboot (in /tmp /var/tmp etc... could gain up to 30 sec? if it is full of garbbage...)

This is not the case with the server. Actually i used to get it in the network within less than 10 sec .Recently our network team changed their switch and we started getting this problem.

I couln't ignore it as one of the service is getting affected due to this(The service is custom build by oracle as a part of Oracle VM 2.2)
They say there is a parameter called up delay (they them self not sure on this) and asking us to change it at server end.

OK..
So how is the network interface configured?
If GB both sides have to be in autodetect
If 100MB then force both sides to FD
Most time it is due to getting the swith port and lan card to agree on speed and HD/FD...

Funny you mention this... I have a server lost in the country side that need rebooting after some crazy power failure test operation where we loose connection... To me the switch takes more time than the server, and the server not finding network believes he is a standalone and so gives up inet configuration... and as you say it used to work till a few months ago

CISCO stuff?

The interfaces are at auto negotiate @both end i.e server and switch.1000gb
There is a delay in creating LACP establishment (@server end its LACP mode 4).
@Switch end they enabled port fast and spanning tree protocol is disabled.


Yes its CISCO stuff