After four years of running Node apps on Linux servers, I have concluded that it could use an OS just for Node. A layer on top of Linux that watched the resources the apps were using and helped you better manage them. It's just a thought at this point.#
Now that I can monitor disk space usage in my ServerMonitor app, I'm getting a better idea of where the problems are. One of the big ones I just discovered is the Dropbox cache. On one of my servers 1/2 of the disk space was used by the cache! The docs say the cache is cleared every three days, but this clearly isn't working on at least one of my systems. According to the docs you can just remove the contents of the ~/Dropbox/.dropbox.cache/ folder. #
A long time ago, in 2010, I was selling my house in Berkeley while living in NYC. On the weekend I had to move out, I had two days to get everything done and then drive back to NY, my main server on EC2 went down. Just died. So I quickly (the movers were arriving) redirected to another server and wrote a post asking if anyone reading my blog had a clue what the problem might be. It never occurred to me that it would be escalated to a problem inside Amazon, that they would treat my post as an urgent request for support. I was asking readers of my blog, many of whom use AWS, if they had any ideas that I could quickly look at that might help me find the problem. This is something I have always done here, today I call them braintrust queries. My readers are smart, generous, know their stuff, and love to show off. I take advantage of that, and am always sure to publicly post the results so it becomes part of the knowledge base of the open web. #- Anyway, I got a call from an engineer at Amazon as a dozen people were working in the house I was turning over in a few hours, loading up stuff, asking for help, etc. The call couldn't have come at a worse time. I explained I couldn't talk, and said goodbye. The guy at the other end sounded surprised and miffed. I imagine in his mind he was doing me a huge favor (he was) and somehow I couldn't stop everything to get the (unexpected) help. At the same time I felt unworthy of the support, yet unable to accept it. #
- To make matters worse, when I finally did find the problem, it was a server log overflowing, getting so large the software just thrashed, it couldn't complete a request. It wasn't Amazon software. As soon as I emptied the log, the server returned to normal.#
- Anyway, I'm often working on stuff that makes their system better. If this is the reason we don't communicate, what a misunderstanding and what a missed opportunity. #