Message boards :
Number crunching :
Unreliable hosts
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 21 Mar 16 Posts: 513 Credit: 4,673,458,277 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I was wondering, since I have a certain host that, let's just say, is a bit unreliable were to spit out some bad WUs, would just that one host be on the server's naughty list or would my entire Username be punished with a lack of work? |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I was wondering, since I have a certain host that, let's just say, is a bit unreliable were to spit out some bad WUs, would just that one host be on the server's naughty list or would my entire Username be punished with a lack of work?This is self defense, not punishment. Only the given host's daily workunit quota gets lower by every failed workunit. When the daily qouta reaches 0, the given host won't be served with another workunit for 24h, then the daily quota is increased by 1, so the host could be served with a workunit. If the given host fails this workunit too, then its daily quota gets reduced to 0, "banning" the host for another 24h. If the workunit is successful, the daily quota is increased by 1, and the host could get another workunit immediately (if there's any in the queue). |
|
Send message Joined: 21 Mar 16 Posts: 513 Credit: 4,673,458,277 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Is there not a system that reads the host's turnaround time and determines which WUs should go to which host? I can imagine that would be pretty useful for priority workloads. |
|
Send message Joined: 2 Jan 09 Posts: 303 Credit: 7,321,800,090 RAC: 245 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Is there not a system that reads the host's turnaround time and determines which WUs should go to which host? I can imagine that would be pretty useful for priority workloads. Seti tried it many years ago for units that had been run unsuccessfully on one host to try and get it validated quicker, it didn't work as the time numbers aren't very reliable. For instance at another project I've been waiting for 5 days for a host to return a workunit where it says the host has a 0.01 days turn around time. Either the numbers have no relation to reality or the host crashed or something, it's a 2 week deadline so I could be waiting for awhile yet!! |
©2025 Universitat Pompeu Fabra