English
I recently started my first Dancer & MongoDB project. MongoDB is my preferred database (at the moment) and Dancer is a great framework, but something is missing: I really prefer using an ORM and didn't find anything for MongoDB which fits my needs.
Did you find a typo in the title? No? You're right. I recently started to love the Perl Dancer Framework - and tried to write my first cronjob for a Dancer based project today.
Look at the title and you're done. No need to add a text to this post if everything is said in the title. Not really? France is too small to hold half of the world's population? No way! My computer told me that 4.294.967.296 people live in France - and computers don't fail, you know?
I've been forced to check the webserver error logs today - and got some really surprising results. A really big amount of requests tried to get /cgi-bin/cache/[some_hex_string] and I suspected our Javascript guys to do something really strange they shouldn't do - but was wrong.
Some weeks ago, I wrote a post about aggregated logging by merging many "UPDATE ... SET count=count+1" SQL statements into one. But there's a problem: All of the data stored in memory is lost if the task or server crashes or if the database dies and doesn't come back until the task exists. Here is how I solved this problem.
Most people would answer "on top of my neck" but that's not always right. Actually, most heads are above their bodies but not every head has a body - which usually is no problem. Trouble starts when someone asks for a head without a body. Headhunters do (but I don't care about them, at least within this post) and web-browsers also do from time to time.
A co-worker asked recently me asked me: "Isn't Perl dead?" It isn't, many people and companies are using it, but no one talks about it.
There are few better ways for taking a SQL server down than SELECT ... JOIN statements. Don't get me wrong - a good JOIN could really fast, but huge JOIN blocks adding 10 or more foreign tables maybe even without using indexes could easily slow down the biggest SQL server. MongoDB has MapReduce to do the job with much less server impact - but more developer brain usage.
Statistics are nice, but may also increase database load. Counting web users (page impressions) may produce a lot of UPDATE requests setting counter=counter+1. I tried to merge them using Gearman.
Vor einiger Zeit war ich auf der YAPC::EU 2012. Einige Talks wurden dort aufgezeichnet und endlich sind die Videos online- wenn auch leider von weniger Talks als erhofft. Hier sind meine Favoriten.
I visited YAPC::EU 2012 this year and some of the talks have been recorded. Finally, the recordings are online - much less than I expected, but here are my favorites out of them.
Perl doesn't care about numbers, integer or float - at least on the script level - but MongoDB does. The driver's default number detection algorithm treats floats as strings and MongoDB doesn't like to compare strings with less-than or greater-than operations.
My development server is a remote Vserver image running under VirtualServer or OpenVZ or anything else. I don't know because I don't care because other people do. I'm used to develop over SSH a lot, this is not problem, but I'm limited to CIFS as a Vserver can't provide NFS. NFS with cachefilesd was fast enough to work on the sources, at least most of the time, but CIFS is a complete fail. Browsing a small directory in my editor may easily take two or three minutes and saving files takes up to 30 seconds.
My laptop has "only" 20 GB for the root-partition / including home, usr and var. This is enough but recent updates dropped my free space below 1 GB and lately below 500 MB. I noticed that various kernel images & modules are wasting spaces but not used at all.
Database load, disk, CPU and memory usage are known ressources on modern servers, but there is another one which is much more important: The I/O volume is limited and can't be raised as easy as all the others. It has no fixed limit and can't be measured (very good). It's a rare (maybe the most rare) good and shouldn't be wasted.
Warnings aren't useless, they may be hints to invisible problems - which could become "big trouble" lateron. Recently, I was trying to track down a warning which didn't make any sense.
