Re: OGR keyserver issues?
"Kevin G. McCoy" <[email protected]>
| Newsgroups | gmane.comp.distributed-net.rc5 |
|---|---|
| Message-ID | <[email protected]> |
On 12/10/2010 4:24 PM, waldo kitty wrote: > On 12/9/2010 12:27, Bruce Slade wrote: >> If I had 50 clients, along with connectivity issues, I'd be running my own >> perproxy client specifically for issues like that and keep it stocked with about >> 3 days work in reserve. > i run less than 50 clients and have always used a perproxy... as for the number > of days of work units it keeps in reserve go, i've let it manage itself these > last several years and have never had a problem, that i know of, with having > available work units for all current tasks for any machine that asks for them :) Bruce, I am not crazy about using personal proxies with a large number of cows. A single point of failure is not a good idea if you are: a) competitive b) have a large number of workstations. The stats/logs generated by a perproxy are next to useless on a very large herd, so there is no advantage there. I am not sure what the perproxy is good for, unless you have a very restrictive firewall and do not have the privileges to modify the rules. I tried using one for a year or two and was completely underwhelmed. I ended up writing my own stats server that watches individual workstation performance. I have my workstations set up to retrieve work individually. Using my stats server I can watch each workstation and if I see a problem developing, I can go fix it to keep my overall key rate up. I have been running dnet on 60-100+ workstations 24/7/365 since the late 90's and am currently #8 in the world on the OGR contest. http://stats.distributed.net/participant/psummary.php?project_id=27&id=17415 <http://stats.distributed.net/participant/psummary.php?project_id=27&id=17415> Best regards, Kevin G. McCoy _______________________________________________ rc5 mailing list rc5-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected] http://lists.distributed.net/mailman/listinfo/rc5