Chandra webserver yawns at slashdot's meager attempts.
The bulk of our traffic is coming from yahoo news.
And while I admit that there were a couple of configuration
issues that were brought to my attention earlier,
they've been resolved, and things are humming along
nicely. Traffic peaked between 8 and 9 (eastern) with
over a million hits in that hour.
If you want the details, we had compiled apache for up to
2048 clients, but had left maxclients set to a meager 512,
which caused some problems up until about 7pm eastern,
when I bumped maxclients to 1536, and watched as actual
connections peaked up around 900. We also had an errant
script that was "gracefully" restarting the web server every
15 minutes, which boosted the load up to around 20
(the server actually didn't seem to mind). Fixed that quick.
The server, by the way is a SunFire 280R (dual 750 MHz) with 4G memory, attached by 100Mbit ethernet (from us
to Harvard is gigabit, and from Harvard to the world is
something really big). Once the errant script was stopped,
load was steady around 1.9 (and I now also realize that
there was an incremental backup in progress since about
6pm).
To paraphrase Kirk:
"I'm laughing at your superior network."
Slashdot Mirror
Please give the original site another try. Minor configuration problems have been resolved, and the server is moving along nicely.
Thank you for your patience.
If you want the details, we had compiled apache for up to 2048 clients, but had left maxclients set to a meager 512, which caused some problems up until about 7pm eastern, when I bumped maxclients to 1536, and watched as actual connections peaked up around 900. We also had an errant script that was "gracefully" restarting the web server every 15 minutes, which boosted the load up to around 20 (the server actually didn't seem to mind). Fixed that quick.
The server, by the way is a SunFire 280R (dual 750 MHz) with 4G memory, attached by 100Mbit ethernet (from us to Harvard is gigabit, and from Harvard to the world is something really big). Once the errant script was stopped, load was steady around 1.9 (and I now also realize that there was an incremental backup in progress since about 6pm).
To paraphrase Kirk:
"I'm laughing at your superior network."