LHC@home

Today the LHC@home project team activated 2 changes in the VM's bootstrap and the CMS Frontier configuration.
The changes enable the use of a local http-proxy, preferably squid, to cache data from cms-frontier.openhtc.io which is heavily used by CMS.

Typical numbers per day and VM (1-core setup):
Requests: 30000-35000
Cache Ratio: 99% in a small cluster running ~30 CMS VMs concurrently


A local proxy should be configured according to the suggestions here:
https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=4611

To use this local proxy just enter it's name (or IP) and it's TCP port into the proxy form of your BOINC client.

Example:
Address: proxy.foo.bar
Port: 3128



Public proxies outside a volunteers LAN should not be used as they are usually not prepared to serve BOINC requests.
This could cause trouble with LHC projects and other projects your BOINC client is attached to.
If you don't have a local proxy openhtc.io will most likely be the fastest external proxy array you can get for LHC@home.


Thanks to Laurence and Ivan

Work fine took a few sec instead of minutes. Speed was much better now.

Glad it works for you. I see some other people are not so happy, but I at least see this as something for, shall we say, more knowledgeable volunteers. It's definitely not for the faint-hearted. :-(

---

The new CMS configuration is working well!

6 CMS tasks connected to cms-frontier.openhtc.io about 160.000 times in total, while the local proxy answered nearly every single request (hitrate > 99%). Additionally, the proxylogs show almost 10.000 connections to cvmfs-stratum-one.cern.ch, where the local proxy answered more than 98%.

Things are looking good, and using a local proxy for CMS tasks seems to be a very good idea.