How many errors are too many?

[poirot1967]

I know the title is vague so I will do my best to describe things. We all strive to get no errors. I take great pleasure when doing a sh int co er, and get nothing but zeros. It makes that 7am cup of coffee taste just that much better. Even when getting a few xmit and rcv or out discards or runt errors, here and there, will not bother me much. When I say a few, I mean less than 20 in a 24 hr period, while for runts and fcs errors, less than 10. But of the errors I listed above, what threshold constitutes a red flag? The previous admin's threshold was 1000. That seems pretty high to me, but being the young gun, my experience pales in comparison.

Is there a generally accepted baseline of the numbers of errors an interface would have on a "normal" day? I am only seeing 3 or 4 interfaces a day that have any errors, and they are attached to proprietary (read finicky) DNA sequencers. The data transfers fine. I am just wondering.

Thanks

Mack

PS Sorry if I didn't thin out the mud enough

[Collin Clark]

I start to 'dig' when I see a couple hundred errors. I have too many devices to monitor on a daily basis, but if I clear counters and then I see them materialize in 10-15 seconds, something is usually wrong. I think a 1000 is too high for fcs, but maybe your wiring is a little freaky or the end devices are finicky or something. Every network is different, but it sounds like your staying on top of it which is good. No best practice, but your learning you network and when something is wrong, you'll know how to recognize it. That's the best thing.