This is an old bug, first discovered in 2003, but now was applied to the hash table functions from the new programming languages. The main idea behind this bug is to create hashes which collide, because collision will eat a lot more cpu cycles on your server. For a post of 2 MB is resulting in a 44 minutes of CPU time which will results in a DOS.

Share this post

Hide Bookmarks

It’s been a while since I configured the latest Dynamic Mass Virtual Hosting Server. Last time I used mod_vhost_alias to create a dynamic virtual hosting and it worked without any problem for what we need in that time. Then we didn’t care about the ftp and virtual users, the sites was updated from web pages and security was pretty much handled by upload application who managed the virtual hosting. Now, the problem is a little bit changed: We need a secure sever which should support ftp virtual users with quota …

Finlay is here. For me, Glassfish 3.1, was one of most awaited piece of software, for all it’s new features and clustering support. Glassfish 3.1 slightly changed the way to install cluster, in a much better and integrated way. This version come with a lot of new functionality, but also with improved parts related to Glassfish 2.1.

Share this post

Hide Bookmarks

Is not a bug, but a pretty annoying configuration thing which appear when you try to configure Enhanced Broker Cluster in Glassfish. I configured my persistent store, for Enhanced Broker Cluster, as jdbc / MySQL:

Share this post

Hide Bookmarks

Why you choose SNMP when Glassfish have strong JMX support? I presume, the answer for all who use SNMP is (almost) the same: Because the architecture of current monitoring solution is not scalable enough, and I cannot load my monitoring servers with supplementary JAVA processes.

Share this post

Hide Bookmarks