Hardware:
The servers running this project are two new 2.6GHz 4MB L2 Intel Dempsey based dual proccessor
dual core nodes, both connected to the internet through a 100Mb connection. The machines are
connected with each other through a direct 1Gb link. Each node has 8GB of RAM on a 1033 MHz FSB
and two 300GB SATA disks. The two SATA disks on each node are setup in a RAID 1 (software)
configuration.
Setup:
Both system are running Scientific Linux 4.3 (RHE4). On node fwnc7128.wks.gorlaeus.net, the MySQL
database is run, which is replicated to node fwnc7129.wks.gorlaeus.net. The URL boinc.gorlaeus.net
is set to both machines and a round-robin load balancing is done between them by the DNS. Because
of that, both machines run the web-services, both can schedule and have access the MySQL database.
Because of a partial NFS setup between the nodes, data is actually shared, but each machine is
capable of taking over in case the other fails. When both machines are up, a load balancing is
achieved in scheduling by running a feeder daemon on each host and by having sepparate
web-services. Odd results are scheduled on fwnc7128, even results on fwnc7129 by the feeders.
Backup:
Both machines reguraly backup the database and the files that have been uploaded or created
independantly.
Homogenous redundancy:
Because each host schedules independantly, both hosts have their own feeder daemon running (with
each a private shmem table of 32MB). Furthermore, because homogeneous redundancy is needed in
this project, it sometimes appears that there is now work. This is not true; usually the work
has already been assigned to other harware, or there is work on the other server.
Status:
The current status of the two servers and which daemons are running where can be obtained by
using the Server Status link on the main page.