At first I didn't really look into it. After all, having the excellent LaunchBar and SpotLight tools handy was making me really happy already. Butler seemed like just another goodie in my menu bar, one among the multitude of tools that try to complement each other, without really making your life that easy at all.
Then I started playing with Butler. First of all, yes, it's a great interface to configure, with a good help section too. The kind of quality we love, and it's even free...
So I played with it, making most of my setups (directories, applications, web bookmarks, even keystrokes macros for my Mail application) invisible from the Finder interface. And I defined some shortcuts to activate all these. Plus shortcuts to rate iTunes songs!
It worked so far very nicely! Sometimes I noticed some delay in the initial caching of Butler, but this will be likely tuned later. Butler is becoming my favorite tool to gain speed in my work. Does what an expensive QuickKeys does. Probably not the most sophisticated of its category but it's really, really great!
Thursday, April 13, 2006
Apache 2.x Weblogic on HPUX
Sometimes, when Apache proxies to a WebLogic server, if the application has an issue and fails in releasing threads (typically you would see in the logs some threads struck and some "Out of Memory" errors), it can have a direct effect on Apache itself. Some child process will top CPU till restarted. Restarting Apache can solve the issue, WebLogic releasing its threads and becoming healthy again.
A fine tuning of Apache and the way the processes and threads are configured is also necessary:
- Apache KeepAlive On (to minimize the number of connections for the same client)
- Apache MaxClients and ThreatdPerChild (one thread = 1 connection)
- TCP parameter tcp_conn_request_max must be set
- TCP parameter tcp_keep_alive_interval (not to have too many sockets in TIME_WAIT state)
- WebLogic managed servers can be allocated some threads with increment (example: start with 25, increment by 5 till 50)
But again the root cause may likely be the WebLogic application itself. So looking into the WebLogic logs is necessary to get the whole picture.
A fine tuning of Apache and the way the processes and threads are configured is also necessary:
- Apache KeepAlive On (to minimize the number of connections for the same client)
- Apache MaxClients and ThreatdPerChild (one thread = 1 connection)
- TCP parameter tcp_conn_request_max must be set
- TCP parameter tcp_keep_alive_interval (not to have too many sockets in TIME_WAIT state)
- WebLogic managed servers can be allocated some threads with increment (example: start with 25, increment by 5 till 50)
But again the root cause may likely be the WebLogic application itself. So looking into the WebLogic logs is necessary to get the whole picture.
Subscribe to:
Posts (Atom)