Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go ยป
No member browsing this thread
Thread Status: Active
Total posts in this thread: 28
Posts: 28   Pages: 3   [ Previous Page | 1 2 3 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 2920 times and has 27 replies Next Thread
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

Msg from Rom Walton, BOINC developer:
Howdy Folks,


We have a new release to test. This release is a general bug fix
release. See the change log for a list of all the fixes.


Noteworthy fixes include:

1. Installation issue on Windows 8.

2. Code-signing support for Mac OSX 10.8.

3. Addition of several ATI GPUs to the detection list.

4. Prevent multiple downloads of app versions for the same app
when the platform differs.

5. Fix menu redraw issues on Linux.

6. Fix crash condition in attach wizard on Linux.

7. Fix log files growing beyond desired size.



Please log test results to http://boinc.berkeley.edu/alpha/ and post bug
reports to this email list.


NOTE: This release is the first release that will install correctly on
Mac OSX 10.8. Mac OSX 10.8 is scheduled to be released next month, so
we need to nail this release down pretty quickly.


Thanks in advance.


----- Rom 29-6-2012 23:27


edit: Version info: [boinc_alpha] BOINC 7.0.30 release for all platforms

--//--

edit2: Oh golly, lucky me, installed it on an ATI fitted device :|
We have pulled 7.0.30 to avoid a bad bug with Nvidia GPUs.

We'll have 7.0.31 released on Monday for testing.

----- Rom 1-7-2012 20:49

----------------------------------------
[Edit 2 times, last edit by Former Member at Jul 1, 2012 7:00:51 PM]
[Jun 30, 2012 9:46:55 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

OK, so 7.0.30 was retracted fast, and 7.0.31 tester is to arrive today. Also put in a request for some additional messaging v.v. types of work fetch rejections. Asked for 4 conditions [main], and we're getting a boatload in the next version, which goes to show that a bit of [one time] persuasion goes a long way in achieving something:
I checked in code to show why work is not being requested in a scheduler RPC.
There are actually quite a few cases:

case CANT_FETCH_WORK_NON_CPU_INTENSIVE:
return "non CPU intensive";
case CANT_FETCH_WORK_SUSPENDED_VIA_GUI:
return "suspended via Manager";(suspended project prevents any work requests)
case CANT_FETCH_WORK_MASTER_URL_FETCH_PENDING:
return "master URL fetch pending";
case CANT_FETCH_WORK_MIN_RPC_TIME:
return "scheduler RPC backoff";
case CANT_FETCH_WORK_DONT_REQUEST_MORE_WORK:
return "\"no new tasks\" requested via Manager";(work fetch has been suspended for this project)
case CANT_FETCH_WORK_DOWNLOAD_STALLED:
return "some download is stalled";
case CANT_FETCH_WORK_RESULT_SUSPENDED:
return "some task is suspended via Manager";(suspended tasks prevent more work requests from this project)
case CANT_FETCH_WORK_TOO_MANY_UPLOADS:
return "too many uploads in progress";
case CANT_FETCH_WORK_NOT_HIGHEST_PRIORITY:
return "project is not highest priority";(project is overworked - other active projects are due computing time)
case CANT_FETCH_WORK_DONT_NEED:
return "don't need";(the client work buffer is full).

-- David

On 01-Jul-2012 2:34 AM, SekeRob wrote:
> On the token of "too many buttons" (one to suspend network *per-project* to
> which I'd not object at all and guaranteed to be forgotten by some it was
> pushed), volunteers do hit buttons and forget or hit buttons they dont
> realise the impact of. Such one is "No new tasks". Time and again we get
> members on the WCG forums where they've hit this work fetch suspend button,
> slipping active memory and when hitting update only see:
>
> 360 World Community Grid 1-7-2012 11:21:05 update requested by user
> 361 World Community Grid 1-7-2012 11:21:08 [sched_op] Starting
> scheduler request
> 362 World Community Grid 1-7-2012 11:21:08 Sending scheduler
> request: Requested by user.
> 363 World Community Grid 1-7-2012 11:21:08 Not reporting or
> requesting tasks
> 364 World Community Grid 1-7-2012 11:21:08 [sched_op] CPU work
> request: 0.00 seconds; 0.00 devices
> 365 World Community Grid 1-7-2012 11:21:13 Scheduler request
> completed
>
> Why oh Why? Well, my KISS answer to this would be, [I'd not be surprised if
> not original, and was so suggested on the WCG forums], to add the reason to
> the zero work request of which I can see 4 main reasons:
>
> (work fetch has been suspended for this project)
> (suspended tasks prevent more work requests from this project)
> (the client work buffer is full).
> (project is overworked - other active projects are due computing time)
>
> Latter though I'm not sure about if working this way in v7. Think the rule
> is now that if there is room in the buffer, to fetch work from the indebted
> project anyhow... the connect not going to waste).
>
> There could be more, but beg for these to be added to sample line 363. They
> will substantially reduce dud support requests.
>
> Thx
>
> -- SekeRob


Of course there will be laments of "too cryptic", but we can work on enhancing these messages later for the layman to understand. At least now a posted log with a "why, oh why" if not clear, will give the knowledgeable folk on the forum more direct indication of which button push was forgotten. A simple translation FAQ will work wonders to help out most, those that are willing to read of course ;O)

--//--

P.S. I've redacted the Dr.Anderson reply and pasted in the 4 *layman* descriptions asked for, as bonus we get 6 extra. Maybe in next [test] client 7.0.32+ :P
[Jul 2, 2012 9:04:06 AM]   Link   Report threatening or abusive post: please login first  Go to top 
johncmacalister2010@gmail.com
Veteran Cruncher
Canada
Joined: Nov 16, 2010
Post Count: 799
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: New BOINC version

I am running 7.0.28, the recommended one...

smile
----------------------------------------


crunching, crunching, crunching.

AMD Ryzen 5 2600 6-core Processor with Windows 11 64 Pro.

AMD Ryzen 7 3700X 8-Core Processor with Windows 11 64 Pro (part time)


smile
[Jul 2, 2012 10:17:58 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

Yeah, right, at WCG? This point release fixed a specific bug for WCG crunchers and left a dozen others. Just *not* WCG material ready for recommendation (Q4-12, maybe, and it will certainly be at the 7.0.3x level if not 7.0.4x by then.) Using and testing and reporting at developers alpha mail list or dev.forum is though much appreciated :D

--//--
[Jul 2, 2012 10:51:31 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

And so 7.0.30 was as quickly retracted as the sun could cross the zenith. Now 7.0.31 out, and behold, the simple exchange of some emails on the alpha mail list with developers and we have....drums rolling... explanations for mortals why work is not being fetched when e.g. the update button is being hit. The change log:
Preliminary Change Log 7.0.30 --> 7.0.31:

Unix build tweak.
Mac client: Update wxMac build script for partial compatibility with OS 10.8 and XCode 4.5
client: when estimating FLOPS for an anonymous-platform app version for which no estimate has been supplied by user, use (CPU speed)*(cpu_usage + 10*gpu_usage) (--> add the 10*)
client: add missing end tag for <pci_info>. Doh!
When the client makes a scheduler RPC without requesting work, and there's a simple reason (e.g. the project is suspended, no-new-tasks, downloads stalled, etc.) show it in the event log. If the reason is more complex, don't try to explain.
client: msg tweak.
client: further msg tweaks.

client: in the job scheduler, there's a check to prevent overcommitting the CPUs if an MT is scheduled. Skip this check for GPU jobs.


Swiftly installed and trying a few scenarios, got these logged:

04/07/2012 10:00:51 | World Community Grid | Not requesting tasks: "no new tasks" requested via Manager

and

04/07/2012 10:02:16 | World Community Grid | work fetch resumed by user
04/07/2012 10:02:27 | World Community Grid | update requested by user
04/07/2012 10:02:31 | World Community Grid | [sched_op] Starting scheduler request
04/07/2012 10:02:31 | World Community Grid | Sending scheduler request: Requested by user.
04/07/2012 10:02:31 | World Community Grid | Not requesting tasks: some task is suspended via Manager

and a more special case

04/07/2012 10:04:16 | Quake-Catcher Network | update requested by user
04/07/2012 10:04:20 | Quake-Catcher Network | [sched_op] Starting scheduler request
04/07/2012 10:04:20 | Quake-Catcher Network | Sending scheduler request: Requested by user.
04/07/2012 10:04:20 | Quake-Catcher Network | Not requesting tasks: non CPU intensive
04/07/2012 10:04:20 | Quake-Catcher Network | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
04/07/2012 10:04:23 | Quake-Catcher Network | Scheduler request completed

Bit cryptic, but it means that you can have only 1 non-intense CPU task at the time for a project , and already having one (Earthquake detection via my laptop in-built accelerometer). [run on the side of a regular full core BOINC task]

So, had you forgotten that you pushed the *No new tasks* button... you'll get told in the event log. It will get quieter on the support forums [in our dreams] ;P

enjoy, or set and forget :D

--//--
[Jul 4, 2012 8:10:20 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

And here a 4th type of report of no new work being send, when hitting "update", or when there is autonomous result reporting and the buffer is full:

930 World Community Grid 4-7-2012 18:21:41 [sched_op] Starting scheduler request
931 World Community Grid 4-7-2012 18:21:41 Sending scheduler request: Requested by user.
932 World Community Grid 4-7-2012 18:21:41 Reporting 1 completed tasks
933 World Community Grid 4-7-2012 18:21:41 Not requesting tasks: don't need
934 World Community Grid 4-7-2012 18:21:41 [sched_op] CPU work request: 0.00 seconds; 0.00 devices
935 World Community Grid 4-7-2012 18:21:46 Scheduler request completed.

The Doctor prefers the KISS texts ;P

--//--
[Jul 4, 2012 4:26:26 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

Msg from Rom Walton, BOINC developer:
Howdy Folks,


We have a new release to test. This release is a general bug fix
release. See the change log for a list of all the fixes.


Noteworthy fixes include:

1. Installation issue on Windows 8.

2. Code-signing support for Mac OSX 10.8.

3. Addition of several ATI GPUs to the detection list.

4. Prevent multiple downloads of app versions for the same app
when the platform differs.

5. Fix menu redraw issues on Linux.

6. Fix crash condition in attach wizard on Linux.

7. Fix log files growing beyond desired size.



Please log test results to http://boinc.berkeley.edu/alpha/ and post bug
reports to this email list.


NOTE: This release is the first release that will install correctly on
Mac OSX 10.8. Mac OSX 10.8 is scheduled to be released next month, so
we need to nail this release down pretty quickly.


Thanks in advance.


----- Rom 29-6-2012 23:27


edit: Version info: [boinc_alpha] BOINC 7.0.30 release for all platforms

--//--

edit2: Oh golly, lucky me, installed it on an ATI fitted device :|
We have pulled 7.0.30 to avoid a bad bug with Nvidia GPUs.

We'll have 7.0.31 released on Monday for testing.

----- Rom 1-7-2012 20:49


Hi.

Anyone know what version for ubuntu has the menu redraw issues on Linux fixed, I've got 7.0.27 that came with 12.04lts.
It worked o.k. yesterday but this morning when i started Boinc up there is only part of the file menu showing so I can't do much with A.T.M. so what is the best way to go, should I just remove it & install an older one or is there a way to fix it.
----------------------------------------
[Edit 1 times, last edit by Former Member at Aug 3, 2012 10:02:39 PM]
[Aug 3, 2012 10:01:16 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

Hi.

Anyone know what version for ubuntu has the menu redraw issues on Linux fixed, I've got 7.0.27 that came with 12.04lts.
It worked o.k. yesterday but this morning when i started Boinc up there is only part of the file menu showing so I can't do much with A.T.M. so what is the best way to go, should I just remove it & install an older one or is there a way to fix it.

Probably in 7.0.29:
http://boinc.berkeley.edu/dev/forum_thread.php?id=6698&nowrap=true#44536

Preliminary Change Log 7.0.28 --> 7.0.29:


MGR: Remove the visuals in the wizard for multi-core and opencl. This area will get a bunch of work in a future commit.
client: show <vbox_window> option in log, and give warning if include this while in sandbox mode.
MGR: Fix build breaks.
MGR: Fix menu redraw issue under Ubuntu's new interface. fixes #1180 (From Huibert)
MGR: Wait until there is valid data in the list box before processing the OnProjectSelected() event.
Fix for linux idle detection bug with USB mice.
client: if available, use GPU model name from CAL for OpenCL model name.
client: added the definitions for the new Windows 7/2008r2 preSP1 and Windows 8/2012 SKUs based on the winnt.h from the Windows 8 RC SKD (also added as proof).
client: added the detection for some more Windows SKU.
client: Updates provided by Teamwork of Planet3Dnow.de to coproc_detect.cpp added CAL_TARGET_ID 21 as : AMD Radeon HD 78x0 series (Pitcairn) (from [P3D] Crashtest).
client, GUI RPC: detect and export the PCI bus, device, and domain #s.
Mac installer: Fix bugs I introduced on May 3 and May 6.
Mac installer: Create or delete Login Items and set screensaver properly when installing for multiple users on OS 10.7, 10.8

[Aug 6, 2012 5:33:45 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: New BOINC version

There is no repository distro at this time that I could find beyond 7.0.27 https://launchpad.net/boinc/+packages which are the easiest installs, better integrating. Since about kernel 3.0.0-21 I've not had the redraw issue with said version [Now kernel 3.0.0-23]. It was a minor issue to me since using BOINCTasks as manager on Windows and Linux [in Wine] gives the superior management experience.
[Aug 6, 2012 7:45:16 AM]   Link   Report threatening or abusive post: please login first  Go to top 
kateiacy
Veteran Cruncher
USA
Joined: Jan 23, 2010
Post Count: 1027
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: New BOINC version

There are two workarounds for the menu redraw issue in Linux.
1. Use the mouse to pull the left side of the Boinc Manager window to the left. The menu reappears (at least on the 7.0.xx versions that I am running).
2. Use Alt key combinations to produce individual menus: Alt-T for Tools, Alt-D for Advanced, etc.
Not optimal, but at least it makes the BOINC Manager usable.
----------------------------------------

[Aug 6, 2012 12:38:59 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 28   Pages: 3   [ Previous Page | 1 2 3 | Next Page ]
[ Jump to Last Post ]
Post new Thread