Boinc 7 has long project backoff

Questions and Answers : Windows : Boinc 7 has long project backoff
Message board moderation

To post messages, you must log in.

AuthorMessage
Mark Lybeck

Send message
Joined: 9 Aug 99
Posts: 245
Credit: 216,677,290
RAC: 173
Finland
Message 1265450 - Posted: 29 Jul 2012, 8:54:50 UTC

Hello,

I noticed that Boinc 7 Has long project backoff for the downloads. Unless you manually hit retry the client retry for download may be easily up to over 5 hours. I do not see a point in having retries in excess of 60 minutes.

Could these retries be configured in the Boinc client by a parameter or otherwise reduced?
ID: 1265450 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1265462 - Posted: 29 Jul 2012, 10:42:33 UTC - in response to Message 1265450.  

If you run SIV on the BOINC Status page you can configure it to do Auto retries,

Claggy
ID: 1265462 · Report as offensive
Mark Lybeck

Send message
Joined: 9 Aug 99
Posts: 245
Credit: 216,677,290
RAC: 173
Finland
Message 1265465 - Posted: 29 Jul 2012, 11:10:15 UTC - in response to Message 1265462.  

If you run SIV on the BOINC Status page you can configure it to do Auto retries,

Claggy


How the hell did someone come to the idea of writig SW like this. It seems that you are able to see and configure things totally unrelated... I mean it seems that if you are in the business of making trucks you also sell fashion-ware... Somehow it is awesome.. but still overkill in a sense....

ID: 1265465 · Report as offensive
Mark Lybeck

Send message
Joined: 9 Aug 99
Posts: 245
Credit: 216,677,290
RAC: 173
Finland
Message 1265466 - Posted: 29 Jul 2012, 11:11:30 UTC - in response to Message 1265462.  

Why not add the auto retry to Boinc instead of SIV?


ID: 1265466 · Report as offensive
Mark Lybeck

Send message
Joined: 9 Aug 99
Posts: 245
Credit: 216,677,290
RAC: 173
Finland
Message 1265467 - Posted: 29 Jul 2012, 11:18:37 UTC - in response to Message 1265462.  

Hello,

Yesterday I actually modified at bit the
http transfersetting to the following

<http_transfer_timeout>90</http_transfer_timeout> <http_transfer_timeout_bps>10000</http_transfer_timeout_bps>

If there is no traffic within a couple of minutes, you better try the next WU.
I do not know if it affects also the scheduler, but I usually get response from the scheduler in 60 seconds so 90 should be enough.


ID: 1265467 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 1265473 - Posted: 29 Jul 2012, 11:43:55 UTC - in response to Message 1265450.  

I do not see a point in having retries in excess of 60 minutes.

Now consider that this project has 220,000+ computers that all try to contact the server at the same time. It'll break the server, it will go off-line as essentially you're doing a denial of service attack on it.

So then consider that this project wants these computers to come back at another time, but please not all at the same time. Spread their load. That's where the back-offs come in. It's to spread the load of all those computers coming to daddy asking for work.

BOINC 6.12 had longer back-offs, up to 12 hours. At least BOINC 7.0 has shorter back-offs.

Stop thinking your computer is alone in this science endeavour. There are some 221,999+ more like yours active here at any given day. With a database that can 'serve' a theoretical maximum of 10,000 connections per second, you should understand why our computers have to be backed-off at times.

And if you just leave things well alone and just let BOINC run to its own devices, after the back-off there may be another (shorter) one, or the work just gets uploaded or downloaded.
ID: 1265473 · Report as offensive
Mark Lybeck

Send message
Joined: 9 Aug 99
Posts: 245
Credit: 216,677,290
RAC: 173
Finland
Message 1265585 - Posted: 29 Jul 2012, 18:26:56 UTC - in response to Message 1265473.  

I do not see a point in having retries in excess of 60 minutes.

Now consider that this project has 220,000+ computers that all try to contact the server at the same time. It'll break the server, it will go off-line as essentially you're doing a denial of service attack on it.
BOINC 6.12 had longer back-offs, up to 12 hours. At least BOINC 7.0 has shorter back-offs.

Stop thinking your computer is alone in this science endeavour. There are some 221,999+ more like yours active here at any given day. With a database that can 'serve' a theoretical maximum of 10,000 connections per second, you should understand why our computers have to be backed-off at times.

And if you just leave things well alone and just let BOINC run to its own devices, after the back-off there may be another (shorter) one, or the work just gets uploaded or downloaded.


Thank you for the clarification. I just got concerned that with Boinc 7.0.28 I had trouble getting WU downloaded faster than my GPU could get them crunched. This was something that was not a problem with 6.10.60.

And when the GPU que was not filled up the CPU que was totally empty. So I had to stagger the Cache starting with 1 day and increase it by one day at a time to be ready for the weekly outage on Tuesday.....


ID: 1265585 · Report as offensive
John McLeod VII
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jul 99
Posts: 24806
Credit: 790,712
RAC: 0
United States
Message 1265748 - Posted: 30 Jul 2012, 3:23:29 UTC

If everything goes smoothly, the backoff is small. It is only when things go wrong that the backoff gets large. IIRC the backoff when things go well, the backoff is 330 seconds.


BOINC WIKI
ID: 1265748 · Report as offensive

Questions and Answers : Windows : Boinc 7 has long project backoff


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.