rnxrx

Similar history here - Pascal was the most common teaching language through the early 90's but by the mid 90's wasn't present in any of the (PC) development shops I encountered, particularly given the popularity of C++ for the Windows ecosystem and the focus given to OOP. Java was still in its infancy in the late 90's and since teaching languages tended to lag behind I'd agree that Pascal -> C/C++ -> Java is about right.

That switch forwards all traffic in hardware. There is no dropping back to CPU processing (for L3 or otherwise). What's more likely is that the t2usd process running hard is actually symptomatic of that hardware periodically reprogrammed, and while the programming is occurring traffic can be dropped. The question is what else is configured on the switch that could be tripping this - again, bad optics? Upstream instability? Flooding CAM tables?

Is it a VPC pair? My suspicion would be that there's something causing a reprogramming event... link flap, STP event, etc. Can you add the full configuration (minus identifying info, obviously) ?

It's important to be clear - anything involving pinging any interface directly on the switch isn't significant. Responding to pings is among the lowest priority for the switch and with a spike in CPU utilization ping responses will be dropped. Is there any possibility of a flapping / bad interface somewhere? Any other surrounding log entries?

Where are you seeing packet loss? A process pinging the switch, or an independent ping through the switch between two independent hosts?

woohoo openwrt

I'm more a network guy by background but I've always kind of liked munin.

It wasn't too bad otherwise, though - and certainly a lot less attenuated than nagios

I seem to recall I had it pulling a lot of data from some heavily loaded 6500's and it managed to completely blitz a pretty hefty VM - to the point where I had to crash the box to rip out the startup

It shit the bed pretty spectacularly when I was working with it.

@MichaelHampton I hear you. I've just seen -so- much money dumped into this stuff over the years and the general maturity level doesn't seem to have advanced as much.

the commercial side has been after the grand unified theory of network management forever, and their packages usually end up taking as much development and customization as just rolling your own

I've been back and forth over the various flavors of Nagios, opennms, zabbix, etc and most are still kind of painful

kind of a rare thing, especially for open source monitoring sw

@ewwhite Yeah.. a lot of stuff I wish it had, but the things they have done they've done well

observium is pretty impressive - especially given how young it is

also check to see if your switch does RMON

go for ntop - generate your own netflow

might be more interesting to put up Netflow top talkers?

ahh... go for stack graphs with a custom cacti template, then

@MarkHenderson What would that visualization even mean? By definition the input traffic on the switch is also output traffic

Also potentially a really neat historical aspect - old-school craftsmanship of rifles from WWI, WW II or even earlier

@WesleyDavid np - oddly enough it turned up on Amazon.. I've put together a pretty comprehensive security system at home but haven't gotten to doing video yet. You are absolutely right that the security guys are pretty weak.

@WesleyDavid dlink.com/us/en/home-solutions/view/network-cameras/… ?

@HopelessN00b We had a guy build a patch cable inventory app for a data center that included gui, relational queries, bar coding, etc.. All Excel, all the time. Six figure contract. Patch cables... Yep.

Woohoo! Excel developer!

or an ISP going out of business

is there something -other- than an oversubscribed ISP?

Fair enough - apologies

Waiting for something existentially satisfying to close vote

Only one vote left

I am not using the Internet, I am using the web!

hadoop clusters of tablets?

@ChrisS It runs over 6a, but the density and power issues associated with the PHY have been horrific. Look at the delta between fiber and copper based GE vs that of 10GE (in density). 48p 10GE line cards in modular chassis are just now becoming available despite 10GE being available in '03...

@MarkHenderson greenfield vs brownfield... If you're sinking $5MM into a new cable plant, you probably want to at least believe it won't be totally irrelevant in 5 yrs.

And 7a will doubtlessly see the light of day... as will 8. As long as people will pay, there will be more.

Behind the scenes even the 10GE stuff has been an absolute nightmare

twisted pair is an absolute nightmare - no way it translates to 40/100 in any sane way

Duh... everyone knows R2 is totally different!

@MichaelHampton Very true. There's a strange line between a thorough taxonomy and breaking things up based purely on volume.

There are RAID questions on unix and serverfault that seem almost identical in nature.

Does anything ever get migrated from unix to serverfault?

The line between unix.se and serverfault seems hazy, but I might be missing something.

I think it would be sort of a shame to split by platform - seeing the Windows and Linux questions together is actually pretty novel.

@HopelessN00b No, I suppose not... but by the same token maybe there should be nginx.stackoverflow.com?

@ChrisS I thought the sharepoint link was a joke too...

notepad.stackexchange.com

That's astonishing. Is there really -that- much that needs to be said about Sharepoint?

@HopelessN00b I don't know if it counts as snark if it's a response to a question involving yellow goo, but I could be wrong.

Next time someone pwns one of those roadside construction signs it needs to include some reference to Gozar the Gozarian