Blog catch up
Krang the content mangement saga
Welp, the major issues with Krang below I haven’t fixed yet with the exception of the ActiveMQ stuff which looks like it’s working fine. I also have a version of Krang using TinyMCE instead of Xinha’s WYSIWYG system. Xinha was causing some issues in regards to people editing and I didn’t waste too much time investigating the actual problem as it seems to have been a reoccurring issue since about 2001 or so. If it hasn’t been fixed by now, it’s either politics or no one cares and with TinyMCE it was working fine. The further improvements don’t make any sense wasting time on because I think New York Magazine needs an entirely new content management system. Fortunately, it seems a majority of people agree. Minus the Krang gripes, the main concern is that it’s hard to create a content management system for any work process without first knowing the process. So the system works like this. Every Friday a new issue of New York Magazine is published. Between that Friday morning and the subsequent Friday there are alot of busy bees. Our backend systems are running on the clustered filesystem, Lustre.
First issue is that Lustre is extremely slow. Any publishing, copy, creating of files takes an inherently long time. This also leads back to Krang and how it does somethings but Lustre is an issue. GFS looks a little faster based on some work of one of my coworkers but all of these filesystems are a pain to administer. I’d like to investigate OpenSolaris with ZFS for all of our filesystem needs. After looking at a couple of demos and realizing that with a terrabyte or two we could simply create our own pools and be done. Need more space? Just add more drives. Backup relevant data and done.
Second issue is that Krang has simply outlived its usefulness; it’s inherently complicated and offers no speed to market at all. Projects that should take a week to complete could end up taking 2-3-4 weeks to fully be completed. We’ve hired two very smart people who I will get up to speed and hopefully Andrew will take off the training wheels when he is ready and deprecate Krang with some of his own ideas. I’d also like to stick my own ideas in regards to the process and help out with some of the coding. Anyway, this is a major blocker for editors, they’d be working primarily on listings (movies, places to go, fashion stuff) and Krang will simply fail. With debug info it shows stuff like race conditions, there is some session time out stuff going on. All sort’s of stuff that Krang simply wasn’t designed to do but extended to do anyway. Sometimes it’s so bad I have to actually restart all of the Krang Services.
Third there is little automation. Anything that comes up more than twice needs to be automated. Anything. One system that needs a tremendous amount of hand holding is in the real estate listings. The system fails, all, of, the, time. The amout of working hours that goes into supporting and just figuring out what is wrong is a complete waste. Even if it’s one hour a week; at the end of the month that’s half a working day dealing with a system that should simply be automated and able to handle it’s own failures. Time wise it’s alot more than that. From investigation part of this politics as there is a middle man that knows the format for X listings and he receives the listings from X people who only deal with him. Wink, wink, pat, pat. Hence alot of time that could be spent fixing other issues is spent back and forth here with this guy.
Fourth all of the systems with the exception of developer server machines should be the same. Right now we are using virtual machines for developer test beds. Which is great, I’m going to build unit tests for all the stuff I work on and urge everyone else to do the same. Anything committed should not fail and should be completely buildable every single night or whenever we want a fresh build for any purpose at all.
Fifth, I spend alot of time dealing with large files. 5-10 gigs moving between machines. I’m thinking about creating a job that automatically dumps data regularly. Part of the problem is that the dev machines should probably be moved from Madison ave to our Data Center. It was probably setup like that because hindsight is 20/20. During the mag move we’ll get this done.
Sixth is user education in regards of where to go when one needs help. There was an issue that involved some people hiring an outside firm who wrote something with Ruby which we don’t support on our production machines. Needless to say alot of companies seem to be fly by night with developers who just bought Ruby on Rails for dummies and have absolutely no clue about anything; at all.
Alot to fix, not enough time; all these issues have to be fixed for things to run smoothly for the mag by Friday. I’ve been really fighting moving into management but it looks like I may have to lobotimize myself and make the jump. The responsibility was offered but I’m clearly not sure if I’m ready yet; I feel I have more direct control over the issues on the ground plus I’m scared i’ll lose my edge on the stuff that i’m trying to fix.
Content Management Systems are flawed
I’m tired right now but I wrote something I’ve got to transcribe on why a majority of Content Management Systems are currently flawed and why. The jist is that current systems need to take into account all aspects of failure and truly become self sufficient systems. Sharing content between people who create it, web serving, file management, user management, search, etc all need to be involved in the process.
Phone
I got myself a BlackBerry 8800 upgrading from the K610i. This wasn’t a necessary upgrade for me personally as it was primarily work related. So far so good. There are alot of things I don’t like about it and it works with Linux in regards to charging just like the older BB’s. Haven’t done any major work with sync’n because seemingly with BlackBerry’s Enterprise server you can wirelessly sync. Which seems to be working out ok for me.
July 4th
Good time, went to a party on the Upper East Side. Was completely and utterly smashed with fatigue. I’m at home now; healing. The fireworks put on by the Macy’s were excellent. The only downside was that as more fireworks exploded in a cacophony of color, the smoke started to block the view. Still got my 15-20 mins worth ooh’s and ahhs. This year had exploding cubes, stars and smiley faces! It was quite impressive.
Work
Been working alot and I don’t even know on what exactly. I’d guess it’s a mixture of all the above. I’m going to take vacation sometime in December, only 6 months to go 
Crossfire
Some idiot knicked my car, have to take Christine into the body shop for repair. I was traumatized over the situation and am still infuriated.