[eclipse.org-committers] Eclipse infrastructure update

["Eclipse WebMaster (Denis Roy)" <webmaster@xxxxxxxxxxx>]

Greetings,

A few words about the state of the Eclipse servers...


The Good
========

Ganymede rolled out with maximum excitement and minimal outages. 
Eclipse.org throughput peaked at 270 megabits/second on release day.

Also, mailing list and newsgroup posts now only take a few minutes to 
appear on the archives, instead of hours, while consuming fewer resources.


The Bad
=======

Our main database/file server is saturated at peak times*.  This is 
affecting performance to some sites, such as the Wiki. We have plans to 
fix this.

* Peak times for Eclipse servers are every morning, Monday-Thursday from 
2:00am to 6:00am and 8:00am to 11:00am Eastern. For some reason, 
Friday's load is much lighter.


The Ugly
========

Disk space usage has peaked for Ganymede, and I'm guessing many teams 
left for vacation without cleaning up dev builds.

- Archive/Download.eclipse.org: 247G and 249G respectively.  Since 
January/08 we've been adding 17G/month.

- Build.eclipse.org:  /shared is at 80% capacity, with only 44G free. 
WTP and EPP are consuming >50G each.

- Tape backup: 3.1T/month, up from 2.5T in April. Tape backup space over 
3.0T/month incurs cost to the Foundation.

Anything you do to reduce disk space waste is a benefit to us all.


Thanks for reading,

Denis



--

Eclipse WebMaster - webmaster@xxxxxxxxxxx
Questions? Consult the WebMaster FAQ at 
http://wiki.eclipse.org/index.php/Webmaster_FAQ