Advanced search

Message boards : Number crunching : Free hard disk for long runs

Author Message
Profile Zarck
Send message
Joined: 16 Aug 08
Posts: 145
Credit: 328,473,995
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41449 - Posted: 30 Jun 2015 | 15:22:54 UTC
Last modified: 30 Jun 2015 | 15:31:03 UTC

131 Go free on my hard disk and i receipt this message... !!!

GPUGRID: Notice from server
Long runs (8-12 hours on fastest card) needs 194.64MB more disk space. You currently have 4573.73 MB available and it needs 4768.37 MB.
30/06/2015 17:21:40

@+
*_*
____________

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41451 - Posted: 30 Jun 2015 | 18:11:57 UTC - in response to Message 41449.

Might this be due to the limits set in Boinc Manager?
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,606,436,851
RAC: 8,676,721
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41452 - Posted: 30 Jun 2015 | 18:23:36 UTC - in response to Message 41451.

Might this be due to the limits set in Boinc Manager?

Or in website preferences - either route can be used, but don't expect to be able to mix them. (Local settings go in as an override file, and lock out future web-based changes).

Profile Zarck
Send message
Joined: 16 Aug 08
Posts: 145
Credit: 328,473,995
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41453 - Posted: 30 Jun 2015 | 22:07:30 UTC - in response to Message 41452.

I have not changed anything, he calculated a long drive, and then the next unit I have this message ... I uninstall and reinstall Boinc no avail ...
____________

klepel
Send message
Joined: 23 Dec 09
Posts: 189
Credit: 4,196,461,293
RAC: 1,617,116
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41454 - Posted: 1 Jul 2015 | 0:04:06 UTC

As you have/are participating in a lot of different projects you might have some crashed tasks, which have not been deleted correctly in your data-directory:
\\ProgramData (something like this as mine is in Spanish)\BOINC\projects

The correct path for you, you will find in your BOINC Manager Data log:
5th line: 26/06/2015 07:43:42 p.m. | | Data directory: D:\[…]\Datos de programa\BOINC

This causes an overflow in your Programm Data and your will hit your “max. set disk space limit”. I have this occasionally with climteprediction.net as each task is accumulating quite a lot of data and if the task crashes the data set is not completely deleted.

I recommend you to check all your sub-projects for obsolete data and delete it.

This should solve your problem. Reinstalling BOINC does not! As BOINC will point to the same folder with the old data sets.

Hope this helps!

Profile Zarck
Send message
Joined: 16 Aug 08
Posts: 145
Credit: 328,473,995
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41455 - Posted: 1 Jul 2015 | 8:35:27 UTC

Gracias por su respuesta a mi problema se resuelve, borré el directorio GPUGRID y unidades GPUGRID se descargan de nuevo.

@+
*_*

Merci pour votre réponse mon problème est résolu, j'ai effacé le répertoire GpuGrid , et les unités GpuGrid sont à nouveau téléchargés.

@+
*_*

Thank you for your reply my problem is solved, I erased the GPUGrid directory, and GPUGrid units are downloaded again.

@+
*_*
____________

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,606,436,851
RAC: 8,676,721
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41457 - Posted: 1 Jul 2015 | 9:29:20 UTC - in response to Message 41455.

If you choose to address the symptom, rather than to diagnose the problem, the simplest way to achieve that is to detach your client from the project, and then re-attach it again.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41459 - Posted: 1 Jul 2015 | 14:00:43 UTC
Last modified: 1 Jul 2015 | 14:01:05 UTC

01/07/2015 15:45:05 | GPUGRID | Message from server: Long runs (8-12 hours on fastest card) needs 1373.40MB more disk space. You currently have 3394.98 MB available and it needs 4768.37 MB.
Well I've received the same error message.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41463 - Posted: 2 Jul 2015 | 8:07:00 UTC

I've received this error message on my other two hosts.
Nothing has changed on these hosts by me, and the free disk space haven't been decreased on them lately that much (by 1300MB, as the log says).
This is a pretty annoying bug, because my spare GPU projects still can crunch on the GPUs.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,606,436,851
RAC: 8,676,721
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41464 - Posted: 2 Jul 2015 | 9:49:53 UTC - in response to Message 41463.

I've received this error message on my other two hosts.
Nothing has changed on these hosts by me, and the free disk space haven't been decreased on them lately that much (by 1300MB, as the log says).
This is a pretty annoying bug, because my spare GPU projects still can crunch on the GPUs.

Well, if you're calling it a bug, I suggest you do some digging into - for example - BOINC's startup reports about what disk space exists on your computer, and what your preferences allow it to use. Post the opening lines from the Event Log here for others to analyse, if you're not familiar with doing that yourself.

For what it's worth, long run tasks are processing normally here, so it isn't a project-wide problem - it's something specific to individual machines.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41465 - Posted: 2 Jul 2015 | 10:43:24 UTC - in response to Message 41464.
Last modified: 2 Jul 2015 | 11:18:56 UTC

Well, if you're calling it a bug, I suggest you do some digging into - for example - BOINC's startup reports about what disk space exists on your computer, and what your preferences allow it to use. Post the opening lines from the Event Log here for others to analyse, if you're not familiar with doing that yourself.

Calling it a bug was the conclusion of my analysis of the event log.

For what it's worth, long run tasks are processing normally here, so it isn't a project-wide problem - it's something specific to individual machines.

Yep, but I have 7 active hosts, and it seems to spread across them. I'm using only 3 of them for other purposes, but it does not interfere with the partition BOINC resides, as I used to boot to Windows 7 when I'm playing.

Here's my event log with my comments in blue:

2015. 07. 02. 9:36:50 | | Starting BOINC client version 7.4.42 for windows_x86_64
2015. 07. 02. 9:36:50 | | log flags: file_xfer, sched_ops, task
2015. 07. 02. 9:36:50 | | Libraries: libcurl/7.39.0 OpenSSL/1.0.1j zlib/1.2.8
2015. 07. 02. 9:36:50 | | Data directory: C:\Documents and Settings\All Users.WINDOWS.0\Application Data\BOINC
2015. 07. 02. 9:36:50 | | Running under account user
2015. 07. 02. 9:36:50 | | CUDA: NVIDIA GPU 0: GeForce GTX 680 (driver version 353.30, CUDA version 7.5, compute capability 3.0, 2048MB, 1996MB available, 3693 GFLOPS peak)
2015. 07. 02. 9:36:50 | | CUDA: NVIDIA GPU 1: GeForce GTX 670 (driver version 353.30, CUDA version 7.5, compute capability 3.0, 2048MB, 2003MB available, 2845 GFLOPS peak)
2015. 07. 02. 9:36:50 | | OpenCL: NVIDIA GPU 0: GeForce GTX 680 (driver version 353.30, device version OpenCL 1.2 CUDA, 2048MB, 1996MB available, 3693 GFLOPS peak)
2015. 07. 02. 9:36:50 | | OpenCL: NVIDIA GPU 1: GeForce GTX 670 (driver version 353.30, device version OpenCL 1.2 CUDA, 2048MB, 2003MB available, 2845 GFLOPS peak)
2015. 07. 02. 9:36:50 | | Host name: Noisy-Cricket
2015. 07. 02. 9:36:50 | | Processor: 12 GenuineIntel Intel(R) Core(TM) i7 CPU 970 @ 3.20GHz [Family 6 Model 44 Stepping 2]
2015. 07. 02. 9:36:50 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 popcnt aes syscall nx lm vmx tm2 pbe
2015. 07. 02. 9:36:50 | | OS: Microsoft Windows XP: Professional x64 Edition, Service Pack 2, (05.02.3790.00)
2015. 07. 02. 9:36:50 | | Memory: 8.99 GB physical, 9.55 GB virtual
2015. 07. 02. 9:36:50 | | Disk: 20.00 GB total, 3.58 GB free
That's correct.
2015. 07. 02. 9:36:50 | | Local time is UTC +2 hours
2015. 07. 02. 9:36:50 | | Config: report completed tasks immediately
2015. 07. 02. 9:36:50 | | Config: use all coprocessors
2015. 07. 02. 9:36:50 | rosetta@home | URL http://boinc.bakerlab.org/rosetta/; Computer ID 1507038; resource share 99
2015. 07. 02. 9:36:50 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7497459; resource share 0
2015. 07. 02. 9:36:50 | GPUGRID | URL http://www.gpugrid.net/; Computer ID 116314; resource share 100
2015. 07. 02. 9:36:50 | GPUGRID | General prefs: from GPUGRID (last modified 23-Sep-2014 19:39:00)
2015. 07. 02. 9:36:50 | GPUGRID | Computer location: school
That's incorrect, as in the preferences this host's location is set to none.
2015. 07. 02. 9:36:50 | GPUGRID | General prefs: no separate prefs for school; using your defaults
2015. 07. 02. 9:36:50 | | Reading preferences override file
2015. 07. 02. 9:36:50 | | Preferences:
2015. 07. 02. 9:36:50 | | max memory usage when active: 8286.11MB
2015. 07. 02. 9:36:50 | | max memory usage when idle: 8286.11MB
2015. 07. 02. 9:36:50 | | max disk usage: 5.78GB
The size of the BOINC folder in the application data is 2.47 GB, so it's ok.
2015. 07. 02. 9:36:50 | | max CPUs used: 7
2015. 07. 02. 9:36:50 | | (to change preferences, visit a project web site or select Preferences in the Manager)
2015. 07. 02. 9:36:50 | | Not using a proxy
2015. 07. 02. 9:36:51 | GPUGRID | Sending scheduler request: To fetch work.
2015. 07. 02. 9:36:51 | GPUGRID | Requesting new tasks for NVIDIA GPU
2015. 07. 02. 9:37:00 | GPUGRID | Scheduler request completed: got 0 new tasks
2015. 07. 02. 9:37:00 | GPUGRID | No tasks sent
2015. 07. 02. 9:37:00 | GPUGRID | Message from server: Long runs (8-12 hours on fastest card) needs 1113.40MB more disk space. You currently have 3654.98 MB available and it needs 4768.37 MB.
Why does the server ask for 4.7GB?
My computing preferences are:

Disk and memory usage default Home School Work Disk: use at most 100 GB 100 GB --- 100 GB Disk: leave free at least 0.001 GB 0.001 GB --- 1 GB Disk: use at most 50% of total 50% of total --- 50% of total

The disk and memory usage preferences in BOINC manager are:
Use at most 100.00 GBytes disk space (0 means no restriction) Leave at least 0.00 GBytes free Use at most 50% of the total disk space Use at most 75% of the swap space

2015. 07. 02. 9:37:00 | GPUGRID | There's no work for the applications you've selected.
2015. 07. 02. 9:37:35 | rosetta@home | Sending scheduler request: To fetch work.
2015. 07. 02. 9:37:35 | rosetta@home | Requesting new tasks for NVIDIA GPU
2015. 07. 02. 9:37:38 | rosetta@home | Scheduler request completed: got 0 new tasks
2015. 07. 02. 9:38:00 | SETI@home | Computation for task 01fe15aa.28708.1471.438086664199.12.56_0 finished
2015. 07. 02. 9:38:02 | SETI@home | Started upload of 01fe15aa.28708.1471.438086664199.12.56_0_0
2015. 07. 02. 9:38:03 | SETI@home | Sending scheduler request: To fetch work.
2015. 07. 02. 9:38:03 | SETI@home | Requesting new tasks for NVIDIA GPU
2015. 07. 02. 9:38:05 | SETI@home | Scheduler request completed: got 1 new tasks
2015. 07. 02. 9:38:06 | SETI@home | Finished upload of 01fe15aa.28708.1471.438086664199.12.56_0_0
2015. 07. 02. 9:38:07 | SETI@home | Started download of 30ja13af.24654.4975.438086664206.12.80
2015. 07. 02. 9:38:12 | SETI@home | Finished download of 30ja13af.24654.4975.438086664206.12.80
2015. 07. 02. 9:38:12 | SETI@home | Starting task 30ja13af.24654.4975.438086664206.12.80_0
2015. 07. 02. 9:42:12 | SETI@home | Computation for task 30ja13af.24654.4975.438086664206.12.80_0 finished
2015. 07. 02. 9:42:14 | SETI@home | Started upload of 30ja13af.24654.4975.438086664206.12.80_0_0
2015. 07. 02. 9:42:19 | SETI@home | Finished upload of 30ja13af.24654.4975.438086664206.12.80_0_0

klepel
Send message
Joined: 23 Dec 09
Posts: 189
Credit: 4,196,461,293
RAC: 1,617,116
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41466 - Posted: 2 Jul 2015 | 14:58:15 UTC - in response to Message 41465.

Your answer on your question is here:

2015. 07. 02. 9:36:50 | | Disk: 20.00 GB total, 3.58 GB free
2015. 07. 02. 9:36:50 | | max disk usage: 5.78GB


And here:
2015. 07. 02. 9:37:00 | GPUGRID | Message from server: Long runs (8-12 hours on fastest card) needs 1113.40MB more disk space. You currently have 3654.98 MB available and it needs 4768.37 MB.


You have limited the disk space, which can be use by all BOINC projects together to 5.78GB, and it seems that 2.126 GB are already used by various BOINC projects. Your problem now is that the new Long WUs demand 4.468 GB, but as you have restricted disc use to 5.78 GB and used already 2.126 GB, BOINC asks for 1113.40MB more disk space, so it can download the WU.

You have two ways to go:
Your either delete not used file in your directory "C:\Documents and Settings\All Users.WINDOWS.0\Application Data\BOINC" or you change your settings in Computer Preference under Memory and Disk Usage the maximum allowed disk space, as well as the percentage of disk space your allow BOINC to use of your Hard Disk. However I do see that your Hard Disk is quite small, and therefore this might not be possible.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41467 - Posted: 2 Jul 2015 | 17:25:08 UTC - in response to Message 41466.
Last modified: 2 Jul 2015 | 17:35:09 UTC

Your answer on your question is here:
2015. 07. 02. 9:36:50 | | Disk: 20.00 GB total, 3.58 GB free
2015. 07. 02. 9:36:50 | | max disk usage: 5.78GB

And here:
2015. 07. 02. 9:37:00 | GPUGRID | Message from server: Long runs (8-12 hours on fastest card) needs 1113.40MB more disk space. You currently have 3654.98 MB available and it needs 4768.37 MB.

You have limited the disk space, which can be use by all BOINC projects together to 5.78GB, and it seems that 2.126 GB are already used by various BOINC projects. Your problem now is that the new Long WUs demand 4.468 GB, but as you have restricted disc use to 5.78 GB and used already 2.126 GB, BOINC asks for 1113.40MB more disk space, so it can download the WU.

You have two ways to go:
Your either delete not used file in your directory "C:\Documents and Settings\All Users.WINDOWS.0\Application Data\BOINC" or you change your settings in Computer Preference under Memory and Disk Usage the maximum allowed disk space, as well as the percentage of disk space your allow BOINC to use of your Hard Disk. However I do see that your Hard Disk is quite small, and therefore this might not be possible.

You're wrong. I forgot to say that I've solved this problem before I've posted about it by deleting unnecessary files from various locations, but I've left the BOINC directory as it were before. So the root of the problem is not the size of the BOINC directory and its size limit, but the total free disk space. The long WUs didn't demand that much free disk space before, and these workunits never use that much (4.468GB) disk space. The problematic "very long" workunits had the longest result files (about 187MB). The GPUGrid folder inside the BOINC\projects folder is about 300MB, the two slots of the GPUGrid app is about 44MB each, so the whole GPUGrid project is using less than 400MB disk space on my hosts. It is quite unnecessary to ask for 10 times more disk space than that. I didn't changed my disk usage preferences for a long time, also there was about 3.5GB free disk space on this host for a long time, so this change came from the project's server.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,606,436,851
RAC: 8,676,721
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41468 - Posted: 2 Jul 2015 | 17:37:16 UTC - in response to Message 41467.

It may be a problem with how certain workunit batches were generated, then. I only have one at the moment: it's e16s74_e13s273f15-NOELIA_ETQunboundx1-0-2-RND4364, and it's been given

<rsc_disk_bound>5000000000

- that's five (decimal) gigabytes. May be an extra zero in there. I'll monitor the other task types as I get them.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41469 - Posted: 2 Jul 2015 | 17:45:35 UTC - in response to Message 41468.
Last modified: 2 Jul 2015 | 17:48:47 UTC

It's the same for GERARD_FXCXCL12_LIG's:

<workunit> <name>e7s1_e4s1f339-GERARD_FXCXCL12_LIG_6118481-0-1-RND6976</name> <app_name>acemdlong</app_name> <version_num>847</version_num> <rsc_fpops_est>5000000000000000.000000</rsc_fpops_est> <rsc_fpops_bound>250000000000000000.000000</rsc_fpops_bound> <rsc_memory_bound>300000000.000000</rsc_memory_bound> <rsc_disk_bound>5000000000.000000</rsc_disk_bound>

5000000000/2^20=4768.37158203125

Jacob Klein
Send message
Joined: 11 Oct 08
Posts: 1127
Credit: 1,901,927,545
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41471 - Posted: 4 Jul 2015 | 1:02:16 UTC

MJH:

So... the work units were created with an incorrect <rsc_disk_bound> parameter? Admins, can you confirm?

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41473 - Posted: 4 Jul 2015 | 8:14:11 UTC - in response to Message 41471.

My understanding is that <rsc_disk_bound>n</rsc_disk_bound> defines the maximum allowable write to disk size for the task.

If the task uses more disk space than this it cannot be written to disk and in that event is killed.

If the app calls on this value as being required disk space then the bug is in the app.

If Boinc has changed something that requires there to be n free bytes available on the disk drive then the bug exists within a Boinc update (not very likely).

If during a run you no longer have ~5GB free disk space available to Boinc then that's where it gets complicated. LAIM doesn't apply to GPUGrid GPU tasks, so they have to be saved to disk, but having it disabled for CPU tasks could also force them to be written to disk and you to run out of disk space. If Boinc stops crunching and you don't have the disk space available then you may lose the task. However, tasks don't normally require ~5GB. Maybe the app is using the n value and saying there isn't enough disk space without first looking at the tasks actual size or trying to save the task to disk and waiting for a failure before cancelling? If that's the case then IMO the bug is in the app. While reducing the n value might be a quick fix (for a new batch) it would still be a rigid cut-off point.

Recent NOELIA tasks also use this same n value (4768MB).
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Jacob Klein
Send message
Joined: 11 Oct 08
Posts: 1127
Credit: 1,901,927,545
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41474 - Posted: 4 Jul 2015 | 12:05:57 UTC
Last modified: 4 Jul 2015 | 12:07:02 UTC

I believe the BOINC server code uses the <rsc_disk_bound>, to say "Well, this task could grow to be THIS size... and if you don't have that much free, then I can't give you this task."

And so, again, if these tasks don't really need 4768 MB of disk space, then the bug is in the server-side configuration of these tasks.

We need an admin to confirm the (probably app mis-configuration) problem.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,606,436,851
RAC: 8,676,721
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41475 - Posted: 4 Jul 2015 | 12:31:03 UTC - in response to Message 41473.

My understanding is that <rsc_disk_bound>n</rsc_disk_bound> defines the maximum allowable write to disk size for the task.

If the task uses more disk space than this it cannot be written to disk and in that event is killed.

If the app calls on this value as being required disk space then the bug is in the app.

If Boinc has changed something that requires there to be n free bytes available on the disk drive then the bug exists within a Boinc update (not very likely).

Remember that BOINC is in two parts - client and server. And the app runs in the client only.

What you have described is mostly client behaviour - BOINC will kill a task if it appears to be using more disk space than <rsc_disk_bound>. I say "appears", because in one recent case left-over files from one specific pre-alpha test project were counted into the totals and caused other tasks to abort when they tried to run.

That isn't the problem here. What is being discussed is the refusal of the server to issue tasks if there a risk that the maximum conceivable disk space used by the application turns out not to be available when the time comes. I think that's a right and proper precautionary principle: we all know how frustrating it is when a task spends many hours crunching to 90%, and then crashes. That's wasteful, and it's good to avoid it if we can.

We know that the server checks the available disk space available to the BOINC client on your computer. I don't know exactly what the comparison is made against, but the size that the admins have declared as the maximum possible working space needed by the task sounds like the right number to me.

What seems most likely is that the admins have thought (even if only subconsciously) "hard disks are big and cheap these days, let's stick an extra 0 on for safety" - to avoid the unexpected crashes we've had at this project recently, when one or other of <rsc_fpops_bound> or <max_nbytes> turned out to be inadequate for a new batch of tasks.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41476 - Posted: 4 Jul 2015 | 13:44:53 UTC - in response to Message 41475.

Its a fact that lots of people use smaller secondary hard drives just for the Boinc data and many people also use SSD's (which are faster but smaller in capacity). Then there are people who build dedicated crunching rigs from spare parts, often using smaller drives.

The amount of hard drive space available (at any given time) also depends on what other projects are being crunched and Boinc settings (which you can change at any time). Projects such as ClimatePrediction have bigger tasks but if you are hooked up to 20 projects the Boinc data folder could be 20GB or more - it's not fool-proof. It helps to clear out projects from time to time (detach, or detach and reattach as Richard suggested).

I thought that <rsc_disk_bound>n</rsc_disk_bound> was to stop leaky apps filling up the drive; a kill switch, rather than being used to determine if there is enough free space to run - free space can change depending on other apps...

In the past there have been tasks that use over 1GB GDDR5 and vary in size during the run. I know it's not the same thing but they might be bigger than 500MB when extracted and stored on the drive, so the extra zero might not be a mistake.

Anyway, the requirements (10GB) and recommendations (20GB) to run Long tasks were defined and shared by the project a couple of years ago:

System requirements Minimum requirements GPU: NVIDIA Kepler GPU (CC3.0) (Geforce 600 series and later) [check list] Nvidia driver series 343 or later CPU: 1.6GHz RAM: 1GB (2GB for Vista, W7, Ubuntu) HDD: 10GB free space, no more than 80% used (very important under windows) Recommended requirements GPU: high-end NVIDIA Geforce GTX 700, 900 series [check list] Nvidia driver series 343 or later CPU: 2.4GHz or higher RAM: 4GB, 2GB+ free for Boinc HDD: 20GB Free space or more, fast drive(s)


If you've been getting away with <10GB then you've been lucky.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41632 - Posted: 7 Aug 2015 | 23:40:09 UTC
Last modified: 7 Aug 2015 | 23:41:11 UTC

The workunits still ask for 5*10^9 bytes (4.656GB) free disk space.

<workunit> <name>e2s4_e1s3f352-GERARD_FXCXCL12_LIG_15347362-0-1-RND2483</name> <app_name>acemdlong</app_name> <version_num>847</version_num> <rsc_fpops_est>5000000000000000.000000</rsc_fpops_est> <rsc_fpops_bound>250000000000000000.000000</rsc_fpops_bound> <rsc_memory_bound>300000000.000000</rsc_memory_bound> <rsc_disk_bound>5000000000.000000</rsc_disk_bound>

Is this really necessary, or is there an extra zero in this last parameter?

Jacob Klein
Send message
Joined: 11 Oct 08
Posts: 1127
Credit: 1,901,927,545
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41633 - Posted: 7 Aug 2015 | 23:44:14 UTC
Last modified: 7 Aug 2015 | 23:46:02 UTC

Any BOINC installation that doesn't have 4.65 GB free to for BOINC to use ....
Likely isn't getting GPUGrid work, because the admins still have this screwed up.

I hope the admins look at this, and care enough to fix it.

I tried to prompt them, twice, earlier in this thread, but no luck.

Profile Zarck
Send message
Joined: 16 Aug 08
Posts: 145
Credit: 328,473,995
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41689 - Posted: 21 Aug 2015 | 12:03:29 UTC

Yo estaba de nuevo insuficiente espacio en disco Mensaje .... he aumentado de 8 a 16 GB en las preferencias de Boinc línea, espero que esto va a resolver el problema ...

I was again insufficient disk space Message .... I increased from 8 to 16 GB in the preferences of Boinc online, I hope this will solve the problem ...

J'ai eu à nouveau le message d'insuffisance d'espace disque.... j'ai augmenté de 8 à 16 Go dans les préférences de Boinc en ligne, j'espère que cela va résoudre le problème...

@+
*_*

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41696 - Posted: 24 Aug 2015 | 19:51:55 UTC - in response to Message 41689.

I increased from 8 to 16 GB in the preferences of Boinc online, I hope this will solve the problem ...

Yep, that's the way to solve this problem. It might appear again, depending on what your other projects do, but then you know what you can do.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41921 - Posted: 2 Oct 2015 | 8:21:21 UTC
Last modified: 2 Oct 2015 | 8:23:37 UTC

The free space on my hosts have run below 5*10^9 bytes again, but this time I've find out its reason: the Google Chrome browser has the bad habit of keeping all of its previous installers (~44MB each) in its installation folder, so the size of this folder is continuously increasing. On one of my hosts it was wasting 1.5GB. I haven't found any setting in Chrome which would limit the size of this folder, so I've manually deleted everything except the last one.
I still think this

<rsc_disk_bound>5000000000.000000</rsc_disk_bound>
is an order of magnitude higher than reasonable.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 6,169
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 42948 - Posted: 7 Mar 2016 | 18:55:17 UTC
Last modified: 7 Mar 2016 | 18:58:34 UTC

Houston, we have a problem!
Now the free space required by the long runs raised to 8*10^9 bytes.

2016. 03. 07. 16:25:19 | GPUGRID | Message from server: Long runs (8-12 hours on fastest card) needs 1064.42MB more disk space. You currently have 6564.98 MB available and it needs 7629.39 MB.

Is this intentional, or just an extra zero?
I've checked it on my working host:

123456789 <rsc_disk_bound>8000000000.000000</rsc_disk_bound>

While the short runs need only 300MB:
12345678 <rsc_disk_bound>300000000.000000</rsc_disk_bound>

Please, reduce the disk space requirement of long runs by an order of magnitude as these tasks don't need 7.45GB disk space.
I could resize the XP partition on my hosts if this much space is needed, but I don't think so.

Post to thread

Message boards : Number crunching : Free hard disk for long runs

//