Unable to communicate with server at night

[bascaro]

Good Morning!
I'm having a problem with my Netware Servers
Every Evening at 18:00 the servers are starting to loose their communication with the other servers...
The connection is used to be established 3minutes later again and that problems arent at work hours!

There are 4 Netware Novell Servers in my Network

3 NW6.5 SP5 Server with eDir 8.8SP1
-ZFB_SVR_R2 (our Main server)
-Abraham
-Bart

1 NW5.1 SP3 Server without eDir (its just for Backup)

Abraham and Bart are not losing the connection to each other.
ZFB_SVR_R2 loses the connection to Bart and Abraham so the same in other directions (Bart and Abraham cant connect to ZFB_SVR_R2 and so its the same in the other Direction)

Another error copied out of the ZFB_SVR_R2's syslog says:
13.09.2006 4.36.39 : SERVER-5.70-3651
Severity = 1 Locus = 14 Class = 2
POLICY MANAGER - (5.00-035): Update Connection License failed. Error # C0001001 (Huetter.isb.zfb.ZFB.)

The connection to the other eDir servers is at that time established so i dont know if that error could be caused by the connect and disconnect error. (always it is a sysadmin user!)

Our Clients have a Problem that their Homedrive(all Networkdrives) is/are lost for serveral seconds (seems that it disappears in that time while the servers cant communicate but there isnt any entry in a SysLog)

Yes we have Cisco switches , but they are configured well. The configuration was checked by cisco employees, yesterday.

At least it seems that our ZFB_SVR_R2 is the problem, but what is wrong and how can i fix it?

 

[terry712]

is anything scheduled to run at 6 o'clock
any crons or any weird jobs that a client has scheduled at that time even from a pc

it's too specific a time

 

[marvhuffaker]

Never trust a Cisco guy. They probably don't know what to look for that is related to your problem. They will always say everything is working. You have to make a case for it and provide evidence that their equipment is breaking.. I've never had a Cisco guy look at something the first time and admit their was a problem.

Marvin Huffaker, MCNE

http://www.redjuju.com

 

[bascaro]

at first, thanks for your answers

@Terry
No, there is not anything scheduled at that time...
there arent any users logged on also.
it seems that the servers always loose their connection but they just log it from 6pm-6am into their syslog
there are from 6pm-6am a few (2 or 3) connection losts so there isnt a chance to say that the syslog is turned off due the work hours...

@Marv
We tryed some things yesterday and got a little result

We have two Catalyst 4006 Switches and two Catalyst 3550 Gateways
one of the switches has the ip pool 10.155.32.0 and one has the 10.155.33.0
The Problem got started as the servers were pluged into the .32 switch and we tryed to upgrade the servers (SP5 and eDir88). In that case it was the absolute chaos... every server lost his connection to everyone excepted to the clients from the .33 switch
two days ago we changed the pool of the servers and the problem is as i pointed it in the first post.

the last thing is our clients in the .33 network dont have any problems with their homedrive and that dont epends from the server's pool...
so things got better since we patched the servers on the .33 switch but there is still one error with our netware server R2:
maybe we have an error on our .32-cisco, but
in every patch case, the ZFB_SVR_R2 loses connection to his mates at night.
so do you know any things that i could do to solve that problem? (expected to reinstall the R2)

so far
bascaro

 

[LawnBoy]

Smells like a switch problem to me. Can you swap the .32 and .33 switches and see if the problem moves?

 

[marvhuffaker]

This is what kills me here... Only because I see this all the time (Please don't take this the wrong way)..

This is what I heard by your description: "We made a change on the switch and everything broke. But we still think its a NetWare problem."

I'm not trying to be cynical, but be realistic, if you change something on the switch, that's where you need to look for your solution. Your Novell servers use standard IP just like workstations and other servers. You should be able to do standard IP troubleshooting to pinpoint the problem

Common NetWare server commands:

PING x.x.x.x
IPTRACE x.x.x.x


Marvin Huffaker, MCNE

http://www.redjuju.com

 

[bascaro]

well it seems that we have to replace the .32 to get a result!
if it is replaced ill talk to you again because i think there is still a netware problem left (at the ZFB_SVR_R2)

thanks

 

[marvhuffaker]

Are your servers configured with both IP and IPX?

You don't need IPX, but if IP only, do you have SLP configured?



Marvin Huffaker, MCNE

http://www.redjuju.com

 

[bascaro]

The Problem is solved
It was the NIC!
The card corrupted one bit and this caused an arp cache poisoning at the clients that are in the same subnet as the servers (the Server sent 1 packet with an other MAC-Adress).
The other clients on the .33 switch are connected over a router to the servers... the router didnt notice about the new mac from the server and the packet was sent correctly

now we're configurating more v-lans to get more collision domains or to get faster such errors solved...

Thanks everyone for help i think noone thought about such a thing ^^