server always busy

[Hookemupman]

Is it just me or is the server always to busy to log on?Or am i doing something wrong?Thanks Jason

 

[admin2]

Announcement

Read the announcement that's at the top of each section. The server speed issues have been resolved and it explains why the server was intentionally down today.

 

[KatDiesel]

They are "Tweaking" the software, most of it is trial and error type stuff, they are working to improve it overall. Are just refering to today the server was busy? If so... thats because the search feature was being modified.

 

[Hookemupman]

Its not just today it seems to me its been the last few weeks that i log on and im getting the servers to busy. I get online only once a day to read TDR and half the time its busy just wander what was up. THanks

 

[admin2]

Server...

After the server being able to scream along just fine last night under a record load we figured we could adjust the search minimum back to 3 characters. We decided to perform the re-index during business hours since most people are at work and not using the Internet at that time.



There was a notice on the boards during the re-index that explained it... . however... . 80% of the way into it the re-index caused the load to get high enough that it couldn't even display the notification. Search indexing take SERIOUS CPU and disk horse-power.

 

[Hookemupman]

Thanks for the follow up. Jason

 

[dan_gilson]

I am wondering what you using for a server. I worked on a couple of super servers a few years back. What type of loads do you see?



If this is sensitive stuff to you then please forget that I asked.



Most of the time the site runs really good. Only a few times that I have not been able to login.

 

[admin2]

Server

Linux:



Dual 1 gigahertz PIII

1. 5 gigabyte RAM

2 32 gigabyte Ultra Wide SCSI hard drives

dual 100 mbit Ethernet cards



The problem isn't CPU, RAM, or hard-drive space. Its a known table locking bug/issue with the MySQL database software. Since the VBulletin software is currently tied to MySQL, all we can do is try to set up an environment that avoids or delays the MySQL issue.



Loads typically stay well under 2. 0 during peak (200-300 HTTPD page requests per second) which is good (2. 0 is 100% on a dual CPU system). When MySQL encounters its bug it gets into a race condition that will eventually shoot the load up to 20+ --- it pretty much forces a reboot.