Throughout the day, even during weekends, Rosetta's Website is VERY slow to respond, e.g. 30sec or more, to the point that browsers will time-out. This makes it difficult to read the forums interactively.

It has been an ongoing problem during the 2 months since I've joined.

I think the problem became less apparent when you switched to 8hr WUs, i.e. fewer concurrent connections.

It's NOT a problem with overall TCP/IP network traffic, because e.g. ping works fine. IMO it's a problem with the Apache webserver running out of server pool resources.

% ping boinc.bakerlab.org
PING boinc.bakerlab.org (140.142.20.103): 56 data bytes
64 bytes from 140.142.20.103: icmp_seq=0 ttl=55 time=16.994 ms
64 bytes from 140.142.20.103: icmp_seq=1 ttl=55 time=16.942 ms
64 bytes from 140.142.20.103: icmp_seq=2 ttl=55 time=16.892 ms
64 bytes from 140.142.20.103: icmp_seq=3 ttl=55 time=16.954 ms
64 bytes from 140.142.20.103: icmp_seq=4 ttl=55 time=16.920 ms
64 bytes from 140.142.20.103: icmp_seq=5 ttl=55 time=16.945 ms
64 bytes from 140.142.20.103: icmp_seq=6 ttl=55 time=16.928 ms
64 bytes from 140.142.20.103: icmp_seq=7 ttl=55 time=16.926 ms
^C
--- boinc.bakerlab.org ping statistics ---
8 packets transmitted, 8 packets received, 0% packet loss
round-trip min/avg/max/stddev = 16.892/16.938/16.994/0.028 ms

% telnet boinc.bakerlab.org 80
Trying 140.142.20.103...
telnet: connect to address 140.142.20.103: Operation timed out
telnet: Unable to connect to remote host


i.e. ...sitting there waiting for > 45sec - 60sec without response -> timeout.

I've tried this test from many different locations in EU and US, to make sure it's not a local problem.

Since you're using "Apache/2.0.52 (Red Hat)" Server at boinc.bakerlab.org, you probably need to tweak the MPM, e.g.

Apache performance tuning

---

Best UFO Resources
Wikipedia R@h
How-To: Join Distributed Computing projects that benefit humanity

Throughout the day, even during weekends, Rosetta's Website is VERY slow to respond, e.g. 30sec or more, to the point that browsers will time-out. This makes it difficult to read the forums interactively.

It has been an ongoing problem during the 2 months since I've joined.

I think the problem became less apparent when you switched to 8hr WUs, i.e. fewer concurrent connections.

It's NOT a problem with overall TCP/IP network traffic, because e.g. ping works fine. IMO it's a problem with the Apache webserver running out of server pool resources.

...I've tried this test from many different locations in EU and US, to make sure it's not a local problem.

I too have noticed this performance degradation, and I do report it to the system Administrator. It does fluctuate over time. There are times when the server responds almost without any delay and other times when it is even slower than you have stated. The worst performance occurs when it is mid evening in Western Europe, and mid morning on the west coast of the US. I am told the the shift to the 2 hour default time for work unit processing has multiplied the data flow from the server by a factor of 4 times wha the norm was only a few weeks ago, and this load is causing some performance problems.

They are working on this issue. Thank you for the report, this may help them a lot.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact

Yes, we're aware of the problem. I'm going to be installing a second webserver frontend today in the hopes of reducing the bottleneck...

---