Network Issues

[njc235]

A broadcast storm caused by a loop is like a snowball. When you reboot your switches you are starting again and until the switches hit 80%+ utilisation, the network will appear to be ok. It's only when you hit around 80 that the storm will be evident. Checked with a couple of our engineers and they say almost certainly a loop.

[tinca492]

Hi

I have had several elderly netgear switches cause issuesbecause of fan failure. Are they running hot?

Is the switch firmware up to date?

+1 for reducing the daisy-chaining by plugging each switchdirect to server switch port to reduce hops

 

[pete]

- What's the CPU usage like on the switches? Low-end Netgears do have a lot of features you can turn on, but they rarely have the horsepower to have everything turned on.

- Can you feed SNMP data from them into something like Cacti or Observium so you've an idea of activity over time? Having utilisation and errors logged elsewhere means they'll survive a reboot.

 

Oh, and +1 on stop daisy-chaining them. While it may not solve the current problem, it'll speed everything else up and make it easier to chop out switches when determining the culprit.

[karldenton]

OK, just an update on this. Its half term this week so plugged the 4 switches on there own Gig connection to the server switch so they are no daisy chained. Had a few issues with 2 switches not working. Turns out the 2 ports I had connected them to on the main switch were not working. I'm thinking the main switch is on its way out.

 

I've since downloaded spice works to monitor usage. Unfortunately it doesn't seem to bring back any data. I have set SNMP up on the switch and spiceworks recognises them but doesn't show any data.

 

I've updated the firmware and enabled STP but yes, the switches are running fairly hot.

Edited February 20, 2017 by karldenton

[tinca492]

Looks like its time to go shopping! As they are all fairly old, do you have the budget to be able to reassess your needs and replace the existing flaky switches with more appropriate/manageable unit(s)?

 

If some of the ports have already gone on a unit its fair to assume that further failure is impending. I really would recomend being ahead of the game and replacing it before it fails. If you explain the potential impact with SLT or the Head then I am fairly sure that they wont want to be without Server access.

[karldenton]

I agree. They are usually understanding about ICT stuff. Will get some costs. I was looking at HP 1810-48g / HP 2510g or something equivalent. Edited February 20, 2017 by karldenton

[Bruce123]

We had the above switch and experienced a similar issue, I am not sure we ever got to the bottom of it. It *may* have been a loop, or some kind of firmware issue. It was many years ago...

 

Thanks,

 

Bruce.

[karldenton]

We had the above switch and experienced a similar issue, I am not sure we ever got to the bottom of it. It *may* have been a loop, or some kind of firmware issue. It was many years ago...

 

Thanks,

 

Bruce.

 

Thanks Bruce. Was that an issue with same Netgear ones or the HP ones I'm looking to buy? Thanks.

[Bruce123]

The Netgear

[LytchettNM]

DON'T RUSH!

There are many options to look at, any switches you replace should give you a minimum of 6 + years service so have a look at the bigger picture before committing to new switches.

Are your current switches mounted at the top of the network cabinet up by the ceiling this seem to be the norm, along with more than one switch mounted above/below another with no air spaces, then the other thing is they will have a door which is keep locked!