Re: [boinc_dev] Fwd: Re: [BOINC/boinc] project preference to set number of used cpu's (#1414)

[Richard Haselgrove]

Yes, indeed. But we should do both.
Protect users from interference with their daily work by finding and 
eliminating disruptive bugs.
But also allowing knowledgeable users to select settings, both when 'in use' 
and also when 'idle', which allow maximum scientific output compatible with 
their known and understood work patterns. 

On Thursday, 17 August 2017, 23:00, David Anderson  
wrote:
 

 There have been many cases where the BOINC client caused a problem,
such as running too many VMs and bogging down the system.
Volunteers then ask for a preference that lets them manually fix the problem.

In such cases it's better to figure out the root cause of the original problem
(e.g. that VM memory usage isn't being accounted correctly)
and change BOINC so it doesn't happen.
BOINC should compute invisibly without fiddling with preferences.

Also, job scheduling and work fetch are intertwined.
The work fetch algorithm does a simulation of job scheduling
in deciding how much work to request for the various resources (CPU, GPU).
Changes to the job scheduling policy (e.g. by new prefs)
typically lead to problems with work fetch (e.g. starved resources).

-- D


 Forwarded Message 
Subject:     Re: [BOINC/boinc] project preference to set number of used cpu's 
(#1414)
Date:     Thu, 17 Aug 2017 21:43:57 + (UTC)
From:     Jord 
Reply-To:     BOINC/boinc 

 

To:     BOINC/boinc 
CC:     Subscribed 



Is this still being asked for?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub 
, or mute 
the 
thread 
.

___
boinc_dev mailing list
boinc_dev@ssl.berkeley.edu
https://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

   
___
boinc_dev mailing list
boinc_dev@ssl.berkeley.edu
https://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

[boinc_dev] Fwd: Re: [BOINC/boinc] project preference to set number of used cpu's (#1414)

[David Anderson]

There have been many cases where the BOINC client caused a problem,
such as running too many VMs and bogging down the system.
Volunteers then ask for a preference that lets them manually fix the problem.

In such cases it's better to figure out the root cause of the original problem
(e.g. that VM memory usage isn't being accounted correctly)
and change BOINC so it doesn't happen.
BOINC should compute invisibly without fiddling with preferences.

Also, job scheduling and work fetch are intertwined.
The work fetch algorithm does a simulation of job scheduling
in deciding how much work to request for the various resources (CPU, GPU).
Changes to the job scheduling policy (e.g. by new prefs)
typically lead to problems with work fetch (e.g. starved resources).

-- D


 Forwarded Message 
Subject:Re: [BOINC/boinc] project preference to set number of used 
cpu's (#1414)
Date:   Thu, 17 Aug 2017 21:43:57 + (UTC)
From:   Jord 
Reply-To: 	BOINC/boinc 
 


To: BOINC/boinc 
CC: Subscribed 



Is this still being asked for?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub 
, or mute the 
thread 
.


___
boinc_dev mailing list
boinc_dev@ssl.berkeley.edu
https://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

[boinc_dev] BOINC and NAT64

[Jord van der Elst]

HI all,

On the BOINC forums a user asks the following:
***
Has anyone tried using BOINC in a NAT64+DNS64 network? I've moved my laptop
to such network (after previously using it in IPv4-only and dual-stack
Internet-connected networks), and nothing seems to work - boincstats BAM
won't connect (fails with *Account manager RPC failed: transient HTTP error*),
projects won't connect (tried LHC@home, RNA World and climateprediction.net,
all fail with *Scheduler request failed: Server returned nothing (no
headers, no data).*)

BOINC 7.6.34 on macOS 10.12.6

Dev guide:
https://developer.apple.com/library/content/documentation/NetworkingInternetWeb/Conceptual/NetworkingOverview/UnderstandingandPreparingfortheIPv6Transition/UnderstandingandPreparingfortheIPv6Transition.html

Long story short:

   IPv4-only servers must be accessed via hostname. The translated address
   from the 64:ff9b::/96 prefix is provided by the DNS64 server (using
   google's 2001:4860:4860::6464)
   IPv4 literals won't work, use high-level APIs that will work with either
   IPv4 and IPv6


***
If anyone has an answer to this, you can answer here and I'll forward, or
you can post it directly to him at
https://boinc.berkeley.edu/dev/forum_thread.php?id=11779

With thanks,

-- Jord van der Elst.
___
boinc_dev mailing list
boinc_dev@ssl.berkeley.edu
https://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.