Rebooting equipment

I only reboot if there's an issue or if it might help with troubleshooting. If you count Windows servers, I reboot when I need to due to updates from Microsoft (and I VPN in from home after hours to do this in order not to disrupt my coworkers).

I was pretty much just talking about switches and routers.    I figured that routers and switch reboots would be beneficial as well, just wanted to see if anyone followed a schedule, or if you just did as needed.  As far as servers go, I reboot our terminal servers and a few of our application servers weekly via task scheduler.  This really helps in the performance, and preempts any issues that we might have during the day when we wouldn't be able to reboot.

Gary, I work in an older building that has frequent power issues, even though I have everything on UPS I like to take advantage of power outages to reboot everything that I can. Other than that I reboot when ever I update firmware, or make major changes to the configurations. I find with some equipment even after changes are made that they are not always dynamic in nature. On the server side I have an older W2K server I reboot every 3 days, and the rest of the servers are rebooted weekly using scheduler.

Hello Dan, thanks for joining in on the discussion.  I like your idea of doing reboots while the power is out and the users wouldn't even be affected as I am from the mindset of being more pro-active instead of reactive.  I'd rather not wait for the routers to flake out before rebooting them.  I'd rather prevent them from flaking out. 

How often?  As little as possible!

While you cannot help it sometimes, that's always a risk, particularly when you're working remotely and don't have a KVM or terminal server to get in again if something goes wrong.

We reboot our servers weekely.  But the switches and routers?  Never.  In the 2+ years I've been in the building the only time I rebooted was when I rebuilt a couple of the network closets with new UPS'.  If the equipment is good quality, they shouldn't ever need a reboot.

-Todd

Most routers are good quality, whether the IOS is up to date is a different matter.  I read that if you have multiple routers with multiple IOS versions, that you could run into trouble from time to time.   

You only need to reboot network equipment if you are experiencing memory/cpu issues, or if some portion of the IOS has become unstable (usually detectable due to memory/cpu issues).  If you want to be proactive, then consider using an snmp monitoring tool and collecting statistics.  MRTG is the simplest tool I can think of for that purpose.  Some tools like groundwork, zenoss, and nagios will let you manage the systems remotely and can be scripted to perform a reboot for you when you reach a critical threshold.

Otherwise, there would be no benefit to rebooting the equipment unless it was failing on a consistently regular basis.  Even with that, you would have to wait for route convergence on the routers, systems to be rediscovered, and switches to refresh their link layer states.  Generally that is not an issue, but if you made changes to your configs and forgot to write them to memory, then OOPS!

There would be a power cost savings benefit to shutting office switches and routers off when not needed though.  If your UPS has a clock/timer feature, that could help save some $$ on an annual basis.

We are currently using What's Up Gold to monitor our network resources.  I know that when it starts to miss pings from our network equipment that it is time to schedule a reboot.  Since starting this thread, we haven't had so much as a hiccup on the WAN.   Bill, of all those tools you mentioned, I have only tried nagios.  If I remember correctly, it works a bit similar to What's Up.  Thanks for chiming in.