switches bring down network every 50 mins

[Duran]

I have 6 3300 switches running 2.68,which at this time is the latest. They are connected using matrix modules and the fibre optic ones.
We had a problem with the switches going down and halting the entire network. I had to reboot them all and then they would work.
I then upgraded the firmware to 2.67 as it was then. 2 wks passed with out problem. Then it happened again, so I then upgraded to the 2.68 firmware. Again no worries for two weeks, but the problem is back.
In one of the switches a sonicwall is plugged in. When the switches go down they appear normal to look at. i have replaced each switch and moved around the matrix modules and swapped then with spares to no avail.

Has anyone out there had this, I don't know what to do now. They are going down every 50 mins or so.

Regards,
Paul

 

[cbratcher]

I have a similar problem, but the frequency is much lower. I use an OfficeConnect Dual Speed Hub & Switch (16 ports each). It seems that every few days between 5:00pm and 6:00pm the switch (and possibly the hub) stop sending packets. Within 5 minutes, it's back up again. (No noticable difference in link lights).

We were having the same problem at around 8:00am, but this was due to poor installation (no ups on the switch & hub). The machines and lights in our warehouse drew enough electricity when powering up to cause a sag for the switch/hub circut.

I've analyzed data at everything from (iso) level 2 and up. I've checked the ups logs and there aren't any sags. The worst part is that we ultimately need new switches anyway (our network was poorly designed from the begining), and upper management doesn't want to fund it unless I can prove that the switch & hub are the problem. Which is hard to do when they refuse to fund the purchase or rental of test equipment.

Luckily (for management) most people are wrapping up what they are doing at 5, and are out by 5:30 (normally avoiding the problem). Too bad it isn't happening at a more productive point in the day, they'd push for a resolution much faster.
Clif Bratcher
Systems Administrator

Publishers Storage and Shipping

 

[Vettejax]

Have you tried sniffing the network to see what type of traffic is on it? Also how are they configured? do you have a map, any speical vlans? Tagging? etc, without a map or any sniffs of the network traffic it's hard to say what is causing the problem. We have run 1100,3300,3900
,CoreBuilders 9400,9300, and 9000 without this type of problem.

Thanks Vettejax

 

[andgroup]

Duran,

I would like to know if this issue was ever solved and whether or not you moved up to 2.69, and whether or not the issue was resolved?

Thanks
John Woods John Woods
Network Engineer (MCSE, CCNP)
Advanced Network Design
jwoods@andgroup.net

http://www.andgroup.net

 

[Duran]

Well, it was a bit of a mission, we replaced a couple of switches which didnt help, replaced the matrix modules and the fibre optic cables and still no difference, they then stayed up for 3 weeks and the problem returned. We started to change the order of the switches and made two separate sets of three which were separated by the FX modules. With no special config (there never was) on the switches except for individual IP's that is all that was done in the end, and they have stayed up since. Unreal. I plug you in, dim the lights
Electric Barbarella !

 

[andgroup]

Duran,

So let me get this straight.

You were using matrix cables?
You removed the matrix cables and connected with switches with fiber modules?

Is this what ultimately solved the problems?

So are you telling me to remove the matrix cables? I have not implemented these switches with the matrix cables yet, but I am doing my homework first, because I have heard of other people having nightmares with these cables, and then I have heard other people say that they work just fine with no problems.

What is your opinion?

John Woods
Network Engineer (MCSE, CCNP)
Advanced Network Design
jwoods@andgroup.net

http://www.andgroup.net

 

[ttming]

It failed after two weeks, look like hardware is not the problem-matric cables should be OK, matrix module should be OK and both of these items do not need software to operate on. If it is OK intially, it should be OK then.

Now how you stacked the 6 units of 3300? I guess you have two stacks which is interlink by fiber optic cable since maximum uints per stack is 4. Fiber patch cord should be OK, so now the probelms must be with the switches or other network equipment such as NIC, faulty patch cord, electrical interference, faulty NIC driver.....

It look like there might be some faulty NIC, faulty transmission UTP links (lateley any changes carried out that mught interferred with data transmission in the UTP cables?)

If the NIC is faulty, it will transmit enought faulty frames that will eventually bring down the whole network after a certain duration. If the NIC is at fault, try to see the number of errors transmitted, well you might have to look at individual port (6 x 24??) to see the error frames.

Since you have upgraded the firmware once, the same problem still exist, upgrade another time to 2.69 and if the problem still persist, look like it is not with the switches itself and you have to start looking external to the switches. Well I hope it work,

Anyway, In my network, I have SuperStack II 9300 as core with 3900, 3300 (firmware 2.69)as edge switches and matrix modules, matrix cable, UPT and air connect (wireless) running and there is no problem so far.

tiong ting ming
technical user
Principal
smjk didndings (A K-12 school)
Malaysia