Welcome to MilkyWay@home

Reducing Workunits to Unreliable Hosts


Advanced search

Message boards : News : Reducing Workunits to Unreliable Hosts
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
mmonnin

Send message
Joined: 2 Oct 16
Posts: 133
Credit: 130,397,138
RAC: 39,345
100 million credit badge2 year member badge
Message 67075 - Posted: 12 Feb 2018, 20:17:02 UTC
Last modified: 12 Feb 2018, 20:19:20 UTC

Other projects straight up ban Hosts. If they never notice zero points or constant errors then they won't notice being banned.

That host is from Gridcoin and it was probably ninja installed on someone else's computer for profit w/o them knowing. Def should be banned. Seen way too many errors from Gridcoin users across all projects.
ID: 67075 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileBeemerBiker
Avatar

Send message
Joined: 18 Nov 08
Posts: 117
Credit: 844,735,502
RAC: 883,207
500 million credit badge10 year member badge
Message 67096 - Posted: 17 Feb 2018, 2:32:20 UTC

What about systems like this wingman that have a turnaround of over 1 day and "inprogress" tasks of over 3000? one of my results is in there. Those tasks were accumulated just in last week it seems. I have .06 day turnaround and average under 200 "inprogress"

Limit on task downloads?
ID: 67096 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 25 Feb 13
Posts: 580
Credit: 68,572,777
RAC: 383,832
50 million credit badge6 year member badgeextraordinary contributions badge
Message 67097 - Posted: 17 Feb 2018, 5:05:39 UTC

There is a limit on the number of workunits you can download and he is way over it. That seems like a bug in BOINC. Not sure what happened there, but is seems pretty rare. If it seems to happen more consistently, I will look into it a bit more.

Jake
ID: 67097 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileBeemerBiker
Avatar

Send message
Joined: 18 Nov 08
Posts: 117
Credit: 844,735,502
RAC: 883,207
500 million credit badge10 year member badge
Message 67098 - Posted: 17 Feb 2018, 15:56:29 UTC - in response to Message 67075.  
Last modified: 17 Feb 2018, 16:11:47 UTC


That host is from Gridcoin and it was probably ninja installed on someone else's computer for profit w/o them knowing. Def should be banned. Seen way too many errors from Gridcoin users across all projects.


If the BOINC version is 7.0.80 then the user knowningly allowed the software to download but does not realize that they are actually mining gridcoins with the coins going here

Host 643627 does not seem to be gridcoin related. I assume it is just stupidity. Never attribute to malice what can be explained by stupidity. FWIW, I agree the host should be banned.
ID: 67098 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileBeemerBiker
Avatar

Send message
Joined: 18 Nov 08
Posts: 117
Credit: 844,735,502
RAC: 883,207
500 million credit badge10 year member badge
Message 67099 - Posted: 17 Feb 2018, 17:22:45 UTC - in response to Message 67098.  
Last modified: 17 Feb 2018, 17:29:00 UTC


That host is from Gridcoin and it was probably ninja installed on someone else's computer for profit w/o them knowing. Def should be banned. Seen way too many errors from Gridcoin users across all projects.


If the BOINC version is 7.0.80 then the user knowningly allowed the software to download but does not realize that they are actually mining gridcoins with the coins going here

Host 643627 does not seem to be gridcoin related. I assume it is just stupidity. Never attribute to malice what can be explained by stupidity. FWIW, I agree the host should be banned.


Yes, is member of team gridcoin but not using grcpool. Found that out too late to edit my post. If on grcpool (like I am) one is not the owner of the project and cannot access the project page to make changes. Items like selecting the proper sub-project are much more difficult as the account manager cannot be used.
ID: 67099 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileCruncher Pete

Send message
Joined: 30 Sep 07
Posts: 16
Credit: 212,704,343
RAC: 5
200 million credit badge10 year member badge
Message 67191 - Posted: 3 Mar 2018, 1:44:22 UTC - in response to Message 66988.  
Last modified: 3 Mar 2018, 1:56:05 UTC

Hey Everyone,

I just tried turning on some options to reduce workunits sent to hosts that return a significant number of errors. If you see any issues, please let me know.

Thank you all for your continued support.

Jake


Jake, I wonder that by turning on some options might also have effected New Clients and or clients that have not retuned a value for some time. Like in my case where I have not contibuted to Milkyway for about a year and most of my machines been upgraded since, than that could be the reason for not receing WU's. My GPU's working OK on other Projects. Note that other users have reported this issue in other threads but have not received a reply to date.

N.B. Please read to forum: Number Crunching: Not able to download work for AMD GPU and further down: Unable to download work for NVIDIA GPU. The troops are restless as they have not received one word from any Admin for days.
ID: 67191 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
bluestang

Send message
Joined: 13 Oct 16
Posts: 85
Credit: 275,790,639
RAC: 359,772
200 million credit badge2 year member badge
Message 67301 - Posted: 4 Apr 2018, 0:11:44 UTC

How about we get more than 80 WUs per machine for reliable machines please? When there's issues we run out of WUs fairly quick. At least double it to start as a test.
ID: 67301 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JH30895

Send message
Joined: 30 Dec 09
Posts: 3
Credit: 73,899,688
RAC: 7
50 million credit badge9 year member badge
Message 67302 - Posted: 4 Apr 2018, 4:08:55 UTC

Starting yesterday, the server is supplying my mac minis with GPU tasks that they cannot process, in spite of my BAM parameters specifying do-not-use-Nvidea-GPU. The only remedy so far is to suspend GPU in the BOINC client, but the server still feeds me a supply of GPU tasks that then just sit there. How can I get the server to honor BAM?
ID: 67302 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilemikey
Avatar

Send message
Joined: 8 May 09
Posts: 2228
Credit: 256,161,578
RAC: 165,971
200 million credit badge10 year member badgeextraordinary contributions badge
Message 67303 - Posted: 4 Apr 2018, 9:44:57 UTC - in response to Message 67302.  

Starting yesterday, the server is supplying my mac minis with GPU tasks that they cannot process, in spite of my BAM parameters specifying do-not-use-Nvidea-GPU. The only remedy so far is to suspend GPU in the BOINC client, but the server still feeds me a supply of GPU tasks that then just sit there. How can I get the server to honor BAM?


Try going into your MilkyWay website settings and turning them off, this may be a time when Bam and MW don't talk to each other.
ID: 67303 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JH30895

Send message
Joined: 30 Dec 09
Posts: 3
Credit: 73,899,688
RAC: 7
50 million credit badge9 year member badge
Message 67306 - Posted: 4 Apr 2018, 13:43:27 UTC - in response to Message 67303.  

I have done that. The problem there is that it is a global setting, so my client that CAN process GPU WUs is now cut off also. This needs to be fixed, not just for my benefit, but also because it is responsible for a large number of failed WUs and the "creation" of unreliable hosts. I have over 700 errors in just a few minutes from machines that were previously processing error free.
ID: 67306 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilemikey
Avatar

Send message
Joined: 8 May 09
Posts: 2228
Credit: 256,161,578
RAC: 165,971
200 million credit badge10 year member badgeextraordinary contributions badge
Message 67310 - Posted: 5 Apr 2018, 11:55:15 UTC - in response to Message 67306.  

I have done that. The problem there is that it is a global setting, so my client that CAN process GPU WUs is now cut off also. This needs to be fixed, not just for my benefit, but also because it is responsible for a large number of failed WUs and the "creation" of unreliable hosts. I have over 700 errors in just a few minutes from machines that were previously processing error free.


Put the pc that can do gpu's in a different venue work, school or home, instead of the default then. I do that all the time when I have different pc's doing cpu or gpu projects. I don't like doing both cpu and gpu workunits from the same project on the same pc, they workunits are different lengths and cause cache problems for me. So I put pc a in the default of no gpu units by yes to cpu workunits, and pc b in home and allow gpu units but no cpu workunits.
ID: 67310 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JH30895

Send message
Joined: 30 Dec 09
Posts: 3
Credit: 73,899,688
RAC: 7
50 million credit badge9 year member badge
Message 67311 - Posted: 5 Apr 2018, 15:12:08 UTC - in response to Message 67310.  

Put the pc that can do gpu's in a different venue work, school or home, instead of the default then. I do that all the time when I have different pc's doing cpu or gpu projects. I don't like doing both cpu and gpu workunits from the same project on the same pc, they workunits are different lengths and cause cache problems for me. So I put pc a in the default of no gpu units by yes to cpu workunits, and pc b in home and allow gpu units but no cpu workunits.


Excellent idea! I have adopted this workaround, and it seems to be the solution to my problem. Thanks.

I still believe that the project's failure to honor BAM's parameters (without warning) may be the root of the "unreliable host" problem. The only reason my failed WU count is in the hundreds, instead of thousands, is because I happened to look at the affected machines on the day of the change. Many machines run unattended and churn out errors unnoticed, until they are identified as unreliable and banned.
ID: 67311 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 5 Jul 11
Posts: 138
Credit: 27,075,696
RAC: 0
20 million credit badge7 year member badge
Message 67401 - Posted: 25 Apr 2018, 10:41:47 UTC

Could we increase the tasks sent to reliable hosts? A GPU only gets an hour's work so BOINC keeps going off to my backup project.
ID: 67401 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : News : Reducing Workunits to Unreliable Hosts

©2019 Astroinformatics Group