JensRantil

Yes, I have multiple servers behind nginx. It's very common.

Huh, didn't know. Thanks for pointing that out!

This essentially became the laser of death the other day and lead to cascading failure which eventually brought down our system. That's why I'm posting this.

Very few people know about this and it's really scary. I'm happy people are voting this up to increase some awareness of this.

Potential workarounds: You might think disabling `proxy_next_upstream timeout` will do, but that will also disable connection timeout retry which is not what you want!

Increasing `proxy_connect_timeout` is not an option, because then you risk filling up too many connections in the nginx instance if the upstream server swallows SYN packets or whatnot.

The real workaround: Use haproxy. Serioously.

Looks like Terraform. https://www.terraform.io

Funny their status page hasn't been updated: http://status.twitter.com

Page doesn't seem to work. I get "undefined". Which are the companies mentioned?

I used to work at a VOIP provider where users started reporting choppy audio. After a week or two we nailed it down to customers that had "call recording" feature enabled. Essentially their calls were being recorded and streamed to an audio file to be accessed later through a web interface. After yet another week of investigations we noticed that disk IO was fairly high on machines that had big customers with call recording enabled for all their endpoints. We drilled the IO issues down to the WAV file format that has a header that needs to be updated for every write to accomodate for the updated length of the recording. This required a lot of disk seeks on spinning disks and unfortunately file flushing could not be disabled.. Switching to a RAW audio format that we post-processed after the call resolved the issue.

This is funny because I am basically about to try and diagnose a _very_ similar issue with a VPN tunnel between a Cisco ASA and AWS. I'm also seeing SYN/ACK being occasionally dropped and TCP connection states ending up in WAIT state.

Swedes, Mother's Day is in three weeks. That said, it's never a bad time to call your mother!

I'm surprised GWT is still used out there. It's slow as hell to compile and JS libraries are so much better and more response nowadays.

Related: sometimes I wish most of my data was in JSON streams so I could simply map and reduce the data using jq (http://stedolan.github.io/jq/), pipes and possibly filemap.

Reminds me of "filemap" - a commandline-like map/reduce tool: https://github.com/mfisk/filemap

If you are an engineer you could join the Sveriges Ingenjörer worker's union to get access to their salary database. There you can filter/group by education, age, experience and lots of more.

It's a good idea, but I would expext most applications vulnerable to not run as root. You would need to be root to patch the bash executable.

Lol, they are not using HTTPS to serve that content. Look who's talking...

I don't see what's interesting with this?

This example wasn't enough? https://github.com/mfisk/filemap/wiki/Examples Hadoop experience is not necessary, but you need to know the basic idea behind MapReduce (https://en.wikipedia.org/wiki/MapReduce).

Have a look at `jq`: https://stedolan.github.io/jq/ It makes working with JSON a breeze.

If your security policy allow putting stuff into Amazon... ;)

mrjob looks nice. If you have a Hadoop cluster. But for "medium sized big data problems" FileMap is a very viable alternative if you have people who know their way around a terminal. Especially if you'd like to have something set up pretty fast. For 500 GB of data setting up Hadoop (steep learning curve; name nodes, jobtrackers, thrift API:s, datanodes, zookeeper and whatnots) is a lot of heavy lifting. Not to mention administering it.

Sure you have Cloudera et al., but I'm still trying to figure out if they really make things easier or worse when it two weeks later comes to figuring out why something is broken, or how to start/install additional Hadoop components.

Yes, FileMap was also extremelly easy to set up.

Parallel is awesome. The only problem I have with it is that it doesn't support data localization. If you have 500 GB of data you don't want to copy it to a machine to run a command on it. Better to have the data stored on your cluster directly, which is what FileMap does.

Why don't you want it from Sony?

Why not just switch to this? http://pro.sony.com/bbsc/ssr/show-digitalpaper/resource.solu...

I wish the article would have stated what encryption he used for his data. Apparently not even GCHQ could crack (or so they say...).

Have a look at Pyramid[1] if you plan a slightly larger project. I've been looking at it for years.

[1] http://www.pylonsproject.org/

Why not just use CSS3 animation?

So, I have this great technical solution to make all guns safe. It's kind of a hacker solution, actually. Also, it will also prevent anyone that hasn't been fiddling with their gun to shoot other people.

The solution is simple. Put an infrared camera on the front of the gun. If there's anything warm in front of the gun it simply wont fire. That will make it impossible to shoot other people and only limitting it to cans etc. Possibly you could also use a real camera and do some image analysis to keep the gun from shooting other sensitive objects, such as gas containers etc. Obviously, the firing of the gun will be locked if you take the camera(s) off.

Okay, I admit it. I'm not being serious. A gun is a weapon. The fact that guns are weapons is the problem. There's no technical solution for that.

The only technical solution that will make a gun safe is one that destroys that weapon.

I don't necessarily think that it needs to be a Forever Project, as long as it lasts for at least 6 months or so.

I've heard many smart people recommend having a bunch of small problems to always think of. I think Richard Feynman recommended it.

Some examples of these kind of mental problems that I've been dealing with over the years:

* How long does it take from the time that the planets are aligned to the time that they are aligned once again. This was a problem I thought of in ninth grade and was the reason why I learn programming at that age on my spare time.

* How does one construct a P2P filesharing application that does not have a central single-point of failure. I thought about this for a year or so in high school. Obviously, I also wanted anonyminity for the users if possible.

* Dreaming about the perfect configuration management.

* Mining common log line patterns and automatically extracting useful information for debugging, surveillance etcetera. How I would scale such an application, what it could be used for etcetera. This actually resulted in https://github.com/JensRantil/disco-slct

* Last year or so; How to develop a web application that uses CQRS, event sourcing and a distributed model. This also involves thinking about how this would enable smooth upgrading of subsystems, scaling of such system and security.

Juggling, guitarr playing, programming, sailing, backpacking, trekking and african drumming.