LED Tells You When Your Server Has Melted


One of the benefits of writing for Hackaday is the ability to command at will a legion of readers to descend on unsuspecting web servers. Most of the time a server can handle the load of thousands of connections. On the rare occasion, though, a server is turned into a pile of slag and dross to the satisfaction of us here at Hackaday and to the ire of admins everywhere.

Checking to see if your server is still running or not isn’t terribly interesting, though. [Eric] thought it would be cool to have a proper physical visualization of how busy his server is, and ended up using a blink(1) USB-controllable LED to display his current server load.

The blink(1) is a small, USB controlled RGB LED that can be used as a universal indicator light. [Eric] had the idea of plugging it in to one of his servers and having the brightness of the LED change in response to the load on the server. He did this with a Python script that queries the Google Analytics API and returns a value from 0 to 255 in response to how popular his server was in the last 10 minutes.

Of course, there’s always a chance Hackaday could Slashdot his server. In that case, the blink(1) glows a steady red, alerting [Eric] to his popularity.

37 thoughts on “LED Tells You When Your Server Has Melted

  1. Neat little device, I can think of all sorts of uses for one. But… apparently they’re $30. That just seems outrageous. I can’t imagine how they’re manufacturing them that would drive up the price like that.

    Oh, if you buy 10 or more they’re only $27 a piece. What a deal. :X

      1. lmao. most flexible and easy to use. It’s a friggin RGB LED. you need to write 3 values over serial. you could even implement polling, where it can ask what the current color is. That’s literally all you need. Everything else can be done in software. If an attiny has enough space for the usb library, that’s all you’d need. Plus an rgb led and a usb connector. If you’re making a pcb, put the usb connector right on the PCB. it should cost $10 (for a nice finished and assembled product). $30 is ridiculous.

    1. Lets see…. $10 for 10 pcbs, which each could at least panelize to 5 boards (so 50 boards total for $.20 per board)….$2 for an ATTINY, $1 for an RGB LED, $1 for a usb connector and some passives…. so $5 in parts to build it yourself….. and how much more flexibility can you get then access to an ICSP header?

        1. Once it is out, it should get cheaper as they make new versions built with the profits of old versions.

          That is what they are SUPPOSED to do, but too many people build one thing and then it gets popular and they just rest on their butts until they get a clone made and then they start to sue people for ALL THE MONEY!

      1. Yes, that’s the standard response when anyone has an issue with an open source project.

        I’m not trying to say they’re awful or price gouging or anything… I think it’s just a lost opportunity for them. It’s like they’re charging a prototyping cost for these little widgets, when they could be selling them for a few dollars if they were in proper production. Maybe there simply isn’t enough demand for them to make that step?

        1. No there probably isn’t a lot of call for them, given that snmp is free to implement and widely supported and can give you a lot more information than a simple light…

          Anyone who actually wants to monitor their server will use snmp.

          1. Well I do think it’s worth something to have blinkenlights on the front of servers. When you have racks upon racks of them, it’s nice to have a red light that shows you which one has the fault. But most server chassis have that already anyway.

          2. Personally, whenever I leave my desk to go for a 150 mile round trip to see the servers and fix one I’ve already got knowledge of what machine needs fixing, exactly what’s wrong with it and what unit of what rack I need to pull a server from.

            Generally I’d identify the server that was broken through snmp polling.
            I’d use out of band access to identify failures and then either arrive onsite myself with replacement parts, meet engineers with replacement parts, we may even just have an engineer sent from hp or whatever without feeling the need to actually stand over them..

            Of course the other thing is…
            Using snmp will allow you much better visualisation than using google analytics, and access to much more information. Also if your server breaks Or site becomes unavailanle this won’t tell you why, whilst a decent snmp polling set up will…

            If the server is laying broken in a rack this light won’t even be on!

  2. $30? Are you kidding me? A USB WiFi adapter is $12 and it has an LED in it. I’m sorry, this is ridiculous!!! You plugging something like this is enough to make me drop my subscription!

          1. A colleague of mine had one of those types. I can’t tell if its exactly the same, but it looks as if.
            I can tell you – its a mess. They work by a little *.exe which you have to provide your pop3 credentials to. It then constantly polls the server to see whether there are new mails available. (Thats why they are only compatible with WIN)

  3. haha. When my project was displayed on hackaday last week, I never suspected I’d get that much traffic. The load was way too high for my 5mpbs home connection. Lucky for me, I had a backup server waiting on Amazon EC2.

  4. Funny. I thought that’s what cactus/what’supgold/any number of any free snmp monitoring software. -some if which will even send you an SMS were for…

    But everyone loves an led. Especially an rgb led :)

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.