Posts by Jim1348

1) Message boards : News : nanoHUB@Home server relocation (Message 541)
Posted 13 Feb 2020 by Jim1348
Post:
Is this a dead project??

Since the website is still up, probably not. But I am beginning to think that it is more than a server relocation.
The large number of errors (due to the large number of apps that must be fixed) may have caused them to rethink the situation.

So we may see a new-improved version, if it does come back. But if there are major changes, I wonder if that would be next semester?
You are free to speculate further. We could give a prize for the best guess.
2) Message boards : Number crunching : Transient upload error (Message 520)
Posted 27 Dec 2019 by Jim1348
Post:
I just attached a machine, and downloaded one work unit; it is a _2.
It is stuck in upload. It looks like others have suffered the same fate.
https://boinc.nanohub.org/nanoHUB_at_home/workunit.php?wuid=6664006

I think I will wait until next week before trying more.
3) Message boards : Number crunching : Transient upload error (Message 511)
Posted 23 Dec 2019 by Jim1348
Post:
I can't upload either, but I think you can still download. That will create a big traffic jam soon.
4) Message boards : Number crunching : Write rate to disk (Message 497)
Posted 17 Nov 2019 by Jim1348
Post:
That should work, though I have never tried tmpfs myself.

But I have used Dataram ramdisk with no problems. You then just need to install the BOINC data folder to the ramdisk that you have created. Primo Ramdisk simplifies that a little by allowing the use of "junction points" that just automatically forward the writes from the original BOINC data folder to the ramdisk. I suppose you could use junction points with Dataram also, since they are included in Windows, but I don't recall trying it.

The writes worry me here. How much they really shorten SSD life I don't know, but don't want to find out.
5) Questions and Answers : Windows : Got Tasks, but fail with "Error while computing" (Message 495)
Posted 16 Nov 2019 by Jim1348
Post:
It appears that VirtualBox is installed correctly. Some of the work that failed on your machine completed successfully on other machines.

It could be your antivirus interfering. Be sure to exclude the BOINC data folder.
Even that might not be enough to prevent the real-time monitoring from interfering.; maybe try a different AV, or turn it off as a test.
I think Windows Defender works OK, and should not interfere.
6) Message boards : Number crunching : Invalid rate. (Message 482)
Posted 3 Nov 2019 by Jim1348
Post:
A server abort is a good idea. But I don't think it is a problem with the apps, they just have to be adapted to run in VirtualBox (and BOINC).
They don't normally run them that way, so they have no way to know what works except send them to us. We are the testers.
7) Message boards : Number crunching : Invalid rate. (Message 478)
Posted 3 Nov 2019 by Jim1348
Post:
They have bad batches, due to the large number of apps they are running from different users.
The admin posted a while ago that they fix them as they find them, but with over 400, it takes a while.

Check your machine to ensure that you have enough memory and are not overclocking, etc.
Otherwise, it is something they have to fix.
8) Questions and Answers : Windows : Message: Postponed: VM job unmanageable, restarting later (Message 470)
Posted 29 Oct 2019 by Jim1348
Post:
It doesn't seem to have done anything. Is there something wrong in the script, or do I need to change something in BOINC to point it to the file, or something else?

It is overly complicated. I am not sure what the brackets and the some of the other entries will do.

Just use this:
<app_config>
   <project_max_concurrent>3</project_max_concurrent>
</app_config>

Remember to save it as an ".xml" file. Do not save it as a ".txt" file and then change the suffix.
Then, place it in the program data folder. Since you are on Windows, that will be:
C:\ProgramData\BOINC\projects\boinc.nanohub.org_nanoHUB_at_home

To activate it, just exit BOINC and restart, or reboot your PC.
There is also a "read config" files option in Boinc Manager, so you may not have to restart it. But I do just to make sure.
9) Questions and Answers : Windows : Message: Postponed: VM job unmanageable, restarting later (Message 461)
Posted 22 Oct 2019 by Jim1348
Post:
That message usually indicates a memory problem of some sort. You have 12 GB, and each work unit usually requires 2 GB, sometimes 3.
So run only a maximum of five work units at a time, since you also have to allow for the operating system, etc.

You can set the "use at most % of the processors" in the BOINC manager accordingly, or use an app_config.xml file (search for it, and I think you will find examples).
10) Message boards : News : Request for feedback (Message 455)
Posted 18 Oct 2019 by Jim1348
Post:
FIX THE VALIDATE ERROR(s)!

It seems that the Windows machines do not validate. A large number of your validate errors have validated OK on Linux machines.
And I have never gotten a validate error that I can see on my Ubuntu machines; only "Error while computing", and not too many of them recently.

So they need to adjust their validator (or app) accordingly.

PS: Are you overclocking? That will kill any science project.
11) Message boards : News : Request for feedback (Message 453)
Posted 16 Oct 2019 by Jim1348
Post:
The current batch of WUs take less that twenty minutes to run on my machine anything longer than that results in error. So I abort the longer WUs so save my computer wasted processing time.

They probably choose a time limit so that the slower machines can finish them. But that wastes time on the faster machines when errors occur.
I think they should shorten the timeout. Catering to the lowest common denominator is probably not the most efficient use of resources.
12) Questions and Answers : Wish list : Science thread (Message 440)
Posted 2 Oct 2019 by Jim1348
Post:
Good points. They have a LOT of submitters and projects. Why not ask each of them to write a short (two sentence) description of what they are doing?
13) Message boards : News : Request for feedback (Message 432)
Posted 22 Sep 2019 by Jim1348
Post:
No, I don't blame the students. It is the applications that have to be fixed; I think it has been said there are over 400 of them, and when errors are encountered they are fixed. That is why the error rate is slowly decreasing. But couldn't they be packaged up into longer units?
14) Message boards : News : Request for feedback (Message 430)
Posted 20 Sep 2019 by Jim1348
Post:
The tasks only run for several minutes on a CPU, in a VM.

I am wondering why they are so short. They are presumably created by students/researchers on a notebook. Why not just crunch them there?
They don't seem long enough to send out.
15) Message boards : Number crunching : Validate error (Message 410)
Posted 21 Aug 2019 by Jim1348
Post:
I haven't used it that much with Windows, but I think it is more sensitive to which version of VirtualBox you are using.
Either 5.1.38 or 6.0.10 might work better.

I usually have few problems with Linux (Ubuntu 16.04) with VB 5.1.38.
EDIT: The error rate is maybe 10% or less, and is going down as they fix their apps.
https://boinc.nanohub.org/nanoHUB_at_home/results.php?hostid=3192

And on Windows, you always have to worry about the anti-virus interfering with it. Be sure to exclude the BOINC Data folder.
Also, you could try disabling the AV for a while.
16) Message boards : Number crunching : Validate error (Message 406)
Posted 21 Aug 2019 by Jim1348
Post:
Though I have noticed that my current work units are taking about 2 GB, so maybe you can get away with less.
They vary a bit in size.
17) Message boards : Number crunching : Validate error (Message 405)
Posted 21 Aug 2019 by Jim1348
Post:
Something has to be done about the high number of invalids at this project!

You have only 3.89 GB of memory. That isn't enough to run even one work unit here (except the small ones).
You should start out with 6 GB for the first one, and then add at least 4 GB more for each additional one.
18) Message boards : Number crunching : Validate error (Message 403)
Posted 20 Aug 2019 by Jim1348
Post:
Seeing "vminfo Error: Unable to connect to system D-Bus (2/3): D-Bus not installed" in just about all my invalids.

I think that is in all of them, even the good ones. It does not indicate an error.
19) Message boards : Number crunching : Validate error (Message 399)
Posted 20 Aug 2019 by Jim1348
Post:
Current running task nanoHUB_at_home 1.12 boinc2docker (vbox64_mt) is hammering the hard disk and swap file very hard.

Yes, I have noticed that. I always use a write cache here, on either Linux or Windows.
https://boinc.nanohub.org/nanoHUB_at_home/forum_thread.php?id=46&postid=158#158

And you will need even more memory for it. I would devote about 1 GB per work unit, with a latency (write delay) of maybe 10 minutes or so. On my 8-core machines, I usually use about 8 GB, though half that would help protect the SSD a lot.

But the work units here are short, which makes it a bit easier to deal with.
20) Message boards : Number crunching : Validate error (Message 396)
Posted 19 Aug 2019 by Jim1348
Post:
Also, the anti-virus might be interfering with the process.


It is possible because of slow my internet connection?

I don't think the speed of the internet connection matters. But be sure to exclude the BOINC data folder from your AV. Even that does not always work, since the AV monitors real-time processes, such as internet communications. Any disruptions might affect the validation process.

I would try a different AV, or temporarily disable it.


Next 20


©2020 COPYRIGHT 2017-2018 NCN