It’s ten days since we announced blogs.perl.org and it still doesn’t work properly.
Our mistake was to launch it on a really underpowered server. It worked fine when there were half a dozen of us testing it out, but when the entire Perl community descended on it to take a look at the site (and to sign up for accounts) the server collapsed under the strain.
All of the problems that I’ve been told about so far have been down to memory allocation issues. The server has 512 MB (corrected from KB – it’s bad, but not that bad!) of RAM and it really needs four times that.
Things have calmed down over the last few days though. I suspect that’s largely because people have lost interest in the site and have given up trying to register. We’ve also switched all of the MT processes to using FCGI. I see that a few people have registered successfully and are posting entries on the site. I salute their persistence.
We have a plan for fixing the situation, however, and that will be put in place over the next week or so. The site will be moving to a new server with a more appropriate hardware configuration. We hope to minimise the amount of downtime whilst this happens.
I can only apologise, once more, for the problems. I hope that once the new server is up and running, you’ll all consider giving blogs.perl.org another try.
I hope it’s obvious, but none of these problems should be taken as a reflection on the quality of Movable Type or the work that Six Apart have done for us on the site. The problems all come down to capacity planning on the part of the people running the project.
Which reminds me – if you’re ever looking for someone to do capacity planning on your web site, I’m not the right person for the job!
Leave a Reply