Posts by Crystal Pellet

1) Message boards : BOINC Manager : Application error on Windows 10 shutdown (Message 108597)
Posted 20 Jun 2022 by Crystal Pellet
Post:
When shutting down my Windows 10 machine and boinc / boinc manager running (iconized) without running tasks,
I get a pop-up with an application error caused by boincmgr.exe
It happened never before, but since I upgraded from 7.19.0 --> 7.20.0 It happens with every shutdown.


2) Message boards : News : Android client version 7.18.1 released. (Message 104995)
Posted 10 Aug 2021 by Crystal Pellet
Post:
7.18.1 Recommended Version, but why

1 10 Aug 09:27:32 Starting BOINC client version 7.18.1 for aarch64-android-linux-gnu
2 10 Aug 09:27:32 This a development version of BOINC and may not function properly
3) Message boards : Android : BOINC Android Client 7.18.1 available (Message 104872)
Posted 25 Jul 2021 by Crystal Pellet
Post:
When suspending a task with tasks 'Ready to start' in queue, a new task starts to run (ok).
When resuming the suspended task the status 'suspended' keeps on the Android's device display,
but infact the status is 'Waiting to run' (I can see remotely in BoincTasks).
Only by opening the task's details on the Android device, you can notice that there's no longer a resume button and
have to guess that this means status is 'Waiting to run'. There is only the abort button, but also no pause button.
Thus locally on the android device a 'waiting to run' task can't be suspended again.
4) Message boards : Android : BOINC Android Client 7.18.1 available (Message 104679)
Posted 1 Jul 2021 by Crystal Pellet
Post:
Please report test results here:
https://boinc.berkeley.edu/alpha/test_form.php
I'm testing version 7.18.0 and have some remarks, but I am not an official alpha-tester (and don't want to be), so can't submit your test_form.
Thus I'll post it here:
- After almost a day the UI for BOINC give no access to menus: tasks, projects, preferences. Somehow (zoom-in BOINC-screen) I got access to the event log, but nothing else.
Thanks to BoincTasks remote access, I could see tasks were still running.
- The mini BOINC-icon in the top phone line told me 'Nothing to do'. Uninstalled and install v7.18.0 again. Issue solved.
- Progress bar is not updated when tasks screen active.
5) Message boards : BOINC client : 7.16.11 available for testing for Windows and Macintosh (Message 100855)
Posted 25 Sep 2020 by Crystal Pellet
Post:
Hi Richard,
I suppose with raw log data, you mean stdoutdae.txt with work_fetch_debug enabled.
4 tasks running now, cause I removed WCG's app_config.xml and restarted BOINC.
I did it that way before with an app_config.xml for WCG and got only 1 task.
For the 3 idle cores, I got after 'project update' the message: "Not requesting tasks: don't need (CPU: job cache full; AMD/ATI GPU: )"
I will finish those 4 tasks and test again with the same conditions.
My re-test will take a while cause the tasks remaining time is about 3-4 hours. I'll speed up it a bit by shutting down a busy VM soon.
I could also send to you the whole stdoutdae to your e-mail when you still are with BT.
6) Message boards : BOINC client : 7.16.11 available for testing for Windows and Macintosh (Message 100853)
Posted 25 Sep 2020 by Crystal Pellet
Post:
Testing is over, these clients have been released to the public as recommended.

Situation: Windows 10, BOINC 7.16.11, 8 threads, allowed 50% of cpu's, 1 thread busy, so nidle = 3, no other work in progress, suspended or waiting. All projects NNW, except WCG
Max concurrent = 8 in app_config.xml and fetch work buffer 864.
Maybe David or Richard can explain why shortfall 0 is used in this situation keeping 3 cpu's idle.
7) Message boards : Questions and problems : LHC@Home on BOINC Android Could not connect (Message 99995)
Posted 18 Jul 2020 by Crystal Pellet
Post:
https://photos.app.goo.gl/hawXcG8Fh3SCCkWE9

I am running BOINC 7.4.53 on Android 9

When I try to connect to LHC@Home I get the error Could not connect.


LHC has no WUs for android as far as I know.

LHC Android applications:

SixTrack
Android running on aarch64 502.05 (android) (beta test) 17 Mar 2020, 16:19:35 UTC 7 GigaFLOPS

sixtracktest (beta test)
Android running on aarch64 502.05 (beta test) 10 May 2019, 9:06:28 UTC 2 GigaFLOPS

Both no work atm.
But what Jord tells: connection fails because of 'certificates outdated problem' with the BOINC play store package v 7.4.53
8) Message boards : Questions and problems : VirtualBox Error on Mac Version of BOINC 7.6.33 (Message 75122)
Posted 5 Jan 2017 by Crystal Pellet
Post:
Another possibility is that the message is not coming from ATLAS, but from LHC@home.

The message is sent (why and how to suppress is under investigation), when SixTrack tasks are requested, but the LHC@home server don't has tasks available.
9) Message boards : Questions and problems : VirtualBox Error on Mac Version of BOINC 7.6.33 (Message 75121)
Posted 5 Jan 2017 by Crystal Pellet
Post:
VirtualBox is not installed
..
I installed VirtualBox hours ago and have restarted BOINC and my machine as well and this error message still persists.

Can you help, please?
.. installed hours ago ..?

Could it be that you had a work request to ATLAS in the time VirtualBox wasn't installed.

Maybe the reason for next flag set in your client_state.xml:

<p_vm_extensions_disabled>1</p_vm_extensions_disabled>

Please follow the checklist Jord already mentioned Checklist Version 2 for Atlas@Home on your PC
10) Message boards : Questions and problems : VirtualBox Error on Mac Version of BOINC 7.6.33 (Message 75097)
Posted 4 Jan 2017 by Crystal Pellet
Post:
Did you enabled virtualization in your BIOS?

Since you have i7-3720, I've often seen with Intel 2nd, 3rd, 4th generations it's disabled by default.
11) Message boards : Projects : Migration of applications from vLHCathome to LHC@home (Message 74286)
Posted 22 Nov 2016 by Crystal Pellet
Post:
All three active applications of vLHCathome (former known as Test4Theory) have been implemented in the old LHC@home project (also known as LHC Classic or LHC 1.0).

LHC@home is now the umbrella project for SixTrack, CMS-, LHCb-, Theory- and in the future ATLAS-, ALICE-simulations and a Benchmark Application.

Therefore account creation on the vLHCathome project is closed.

LHC@home consolidation
Migration to LHC@home
12) Message boards : BOINC Manager : BOINC Manager connection and work issues (Message 66793)
Posted 12 Jan 2016 by Crystal Pellet
Post:
New system time (1452329131) < old system time (1452415526); clearing timeouts

When new system time Sat, 09 Jan 2016 08:45:31 GMT (1452329131) is in the past
compared to old system time (1452415526) Sun, 10 Jan 2016 08:45:26 GMT as it's here the case, BOINC will wait until new time is equal to old time.
13) Message boards : BOINC client : BOINC 7 development discussion thread. (Message 65551)
Posted 22 Nov 2015 by Crystal Pellet
Post:
Setting <dont_lower_client_priority> to 1 in cc_config causes the impossibility to change priority of the processes started by boinc.exe like
sciences.exe, wrappers.exe and vboxwrapper.exe, although the user is the owner of the process, so no service install.

I suppose this is an unwanted site effect of the implementation of this tag.


Edit: Not reproducible by restarting BOINC. Probably only happens after starting BOINC Manager at the end of BOINC install procedure when ticked.
14) Message boards : Questions and problems : Increased HDD I/O Activity w/ Version 7.6.(x?) (Message 64392)
Posted 21 Sep 2015 by Crystal Pellet
Post:
(BTW, the VBox activity complained of consists of re-reading a log file, to see if it's changed. Usually it hasn't, so in most cases - except on the oldest of drives - it's probably being read from cache anyway. Don't most HDD activity indicators signify activity on the interface, rather than in the internal mechanics?)

Hello Richard,

In my opinion that vbox.log from the slot directory is written every second with ~32kB disk-i/o's per second.
I think it's copied every time from slots/X/boinc_xxxxxxxxxx/Logs/vbox.log to avoid a lock when reading from that last vbox.log directly.
You also can see that the size of the vbox.log in the slot is flipping regularly between 0 kB and ~64kB.
15) Message boards : Questions and problems : virtualbox 5.0 not recognized by boinc (Message 63034)
Posted 14 Jul 2015 by Crystal Pellet
Post:
I discovered one issue in my former mentioned setup:
..
..
..

Does that always happen when a new job is received? Or only after restoring from a checkpoint?

With every new BOINC-task received and app_config configured as described, the 2-core VM is created and booting, but unusable.
Checkpoints (I suppose in this case you mean snapshots) aren't made.

Selected 2 of those problem tasks returned by ending gracefully:
http://lhcathome2.cern.ch/vLHCathome/result.php?resultid=5044490
http://lhcathome2.cern.ch/vLHCathome/result.php?resultid=5045098
16) Message boards : Questions and problems : virtualbox 5.0 not recognized by boinc (Message 63026)
Posted 13 Jul 2015 by Crystal Pellet
Post:
I discovered one issue in my former mentioned setup:

Configuring the VM with 2 cores by adding <avg_ncpus>2.0</avg_ncpus> in the app_config.xml (what was working fine in VBox 4.3.28 with VBoxwrapper 26167) results in:

- VM created with 2 cores
- VM booting
- During boot phase stops somewhere, although VBoxHeadless.exe keeps on running using a full core.
- Not doing real work.

Maybe this message has something to do with it:

00:01:19.883176 ERROR [COM]: aRC=E_FAIL (0x80004005) aIID={4afe423b-43e0-e9d0-82e8-ceb307940dda} aComponent={MediumWrap} aText={Unknown exception
00:01:19.883176 F:\tinderbox\win-rel\out\win.amd64\release\obj\VBoxAPIWrap\MediumWrap.cpp[2245] (long __cdecl MediumWrap::GetEncryptionSettings(unsigned short **,unsigned short **))}, preserve=false aResultDetail=0


My own Linux-VM with 4 cores on my Windows host is running fine with VirtualBox 5.0, like the projects VM with 1 core is doing OK too.
17) Message boards : Questions and problems : virtualbox 5.0 not recognized by boinc (Message 63019)
Posted 12 Jul 2015 by Crystal Pellet
Post:
I did a test with combo Win7 64bit, vboxwrapper 26169, VirtualBox 5.0, BOINC 7.4.18 and project VirtualLHC@home.


Repeated same test with BOINC version 7.6.3.

Result: http://lhcathome2.cern.ch/vLHCathome/result.php?resultid=5044094
18) Message boards : Questions and problems : virtualbox 5.0 not recognized by boinc (Message 63014)
Posted 12 Jul 2015 by Crystal Pellet
Post:
I build Vboxwrapper 26169 yesterday, which includes the needed changes to support VirtualBox 5.0.

You can find them here: http://boinc.berkeley.edu/trac/wiki/VboxApps

Thanks Rom!

I did a test with combo Win7 64bit, vboxwrapper 26169, VirtualBox 5.0, BOINC 7.4.18 and project VirtualLHC@home.

Result: http://lhcathome2.cern.ch/vLHCathome/result.php?resultid=5043240

Vbox_job.xml:

<vbox_job>
 <os_name>Linux26_64</os_name>
 <memory_size_mb>1024</memory_size_mb>
 <enable_network/>
 <enable_floppyio/>
 <enable_remotedesktop/>
 <enable_cern_dataformat/>
 <job_duration>86400</job_duration>
 <minimum_checkpoint_interval>60</minimum_checkpoint_interval>
 <pf_guest_port>80</pf_guest_port>
 <pf_host_port>7859</pf_host_port>
 <disable_automatic_checkpoints/>
 <enable_vm_savestate_usage/>
</vbox_job> 

I made a manual snapshot in between (task suspended, LAIM off).
Job ran fine and everything cleaned up after the finish.
I've stderr.txt, VBox.log's and VBoxStartup.log available, if interested.
19) Message boards : Questions and problems : Is really Boinc run in idle priority? I think it is not. (Message 62356)
Posted 28 May 2015 by Crystal Pellet
Post:
...
Project science applications for the GPU, (undefined) coprocessor and for VirtualBox run in Below-Normal priority because when these are run at low priority, anything else running at the same time will take away cycles and cause these tasks to unload from memory. GPU, coprocessor and VBox tasks do not remain in memory when BOINC pauses with the Leave in memory preference setting.

Sorry Jord,

But since, I think it was VBox version 4.3.14 Oracle decided for security reasons to create VirtualBox processes more sandboxed.
Since then the Vbox processes are running at 'normal' priority.
BOINC cannot change that, neither the user can (at least for Windows).
20) Message boards : BOINC client : BOINC 7 development discussion thread. (Message 58406)
Posted 3 Dec 2014 by Crystal Pellet
Post:
David Anderson wrote:
There was a bug involving zero-share projects.
I fixed it, and things seem to work, at least in the simulator: http://boinc.berkeley.edu/dev/scenarios/126/simulations/5/timeline.html

We'll port this to 7.4 and do a new release.

-- David

v7.4.32 is now available on the 'download all' page. It installs and runs OK (64-bit Windows), but I don't use the backup project setting. Any testers?

Hi Richard,

I would love to test it, but WCG sent monster BETA's to my machines containing 1 single job in a Workunit, so no checkpointing and running now 14 hours and 50% progress.
Will see tomorrow ;-)

CP

Done! I only tested this issue and can confirm that it's fixed. I set resource share zero for PG before and updated the project with the old client. With 1 idle core:

39 03 Dec 09:53:17 Version change (7.4.18 -> 7.4.32)
87 PrimeGrid 03 Dec 09:53:45 update requested by user
88 PrimeGrid 03 Dec 09:53:46 Sending scheduler request: Requested by user.
89 PrimeGrid 03 Dec 09:53:46 Requesting new tasks for CPU
90 PrimeGrid 03 Dec 09:53:49 Scheduler request completed: got 1 new tasks


Next 20

Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.