[Systems] aslo update.

[dfarning at gmail.com]

It looks like aslo is setteling into three related discussions  aslo at gnaps, aslo at RIT, and aslo via puppet.

1. Gnaps is the data center which houses sunjammer and treehouse.  This pair of machines is in production and easily handling the current load:
http://sunjammer.sugarlabs.org/munin/sugarlabs.org/aslo-web.sugarlabs.org-cpu.html
http://sunjammer.sugarlabs.org/munin/sugarlabs.org/sunjammer.sugarlabs.org-cpu.html
The database is running on sunjammer while 99% of the php load is going to also-web on treehouse.  The one percent remaining on sunjammer is just to keep the caches somewhat warm.

The biggest remaining hole is the data generation scripts.  Currently we are only tracking the pages and downloads served via sunjammer.  We are either going to have to figure out how to parse the loadbalancer logs or parse both sets of apache logs.

We are not losing statistics.  I was just worrying we had changed enough stuff just bring the loadbalancer and additional webnode on line.  We can go back and parse the webnode's logs when we fix the parser.

2.  aslo @ RIT. we are still working with prof. Jacobs and Remy to house the machines in the student innovation center.  I hope we learn more about that tomorrow.

This afternoon alsroot did some final work to bring up http://aslo1.sugarlabs.org/en-US/sugar/ ! . This leads directly to point 3.

3. aslo via puppet.  We have been working the last couple of weeks to do the plumbing so that we can do the configuration management of the cluster via puppet.  Aslo1 (above) was brought online from an initial ubuntu9.10 server install to configured aslo node without anyone ever directly touching the /etc directory.  Full disclaimer:  We cheated by populating the database with a mysql dump from sunjammer and I got lazy earlier tonight and hand edited php.ini.

This gives us a couple of tasks over the next few days.  Hopefully dogi, silbe, or alsroot will try to bring up also2 and aslo3 to prove it works:)  Concurrently we have to:

a. Setup the loadbalancing and HA for the php servers.  For the first run we will probably just use mod_proxy to match what we have working on sunjammer.  We can replace that with squid as needed.

b. Set up the HA and master-master configuration for the mysql server.

c. Set up the log parsers and other cron jobs.

d. Set up the shared file system.  This is the part that scares me the most.  File systems with true replication get complicated fast.  My guess is that we will end up using a combination of nfs, rsync, and heartbeat in the short term.

david
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sugarlabs.org/private/systems/attachments/20100308/8a839420/attachment-0001.htm 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 271 bytes
Desc: OpenPGP digital signature
Url : http://lists.sugarlabs.org/private/systems/attachments/20100308/8a839420/attachment-0001.pgp