Three Linux Processes for One WU?

forloop32
forloop32
Joined: 19 Mar 05
Posts: 7
Credit: 725,230
RAC: 0
Topic 188722

I am new to BOINC. I was very surprised to see how slowly my Linux system seemed compared to others, so I started looking around. I noticed two key things that concern me.

First, on the website I see that I have two results in progress. Having a single CPU system, I expected only one.

Second, I see that there are three Einstein processes running, all with the same parameters. The boinc process starts the first, which starts the second, which starts the third.

Does all of this make sense?

Thanks!

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 4,888
Credit: 29,325,491,074
RAC: 35,237,105

Three Linux Processes for One WU?

> Does all of this make sense?

Unfortunately, no. I don't really understand what you are asking and I don't think you have a correct idea of what is going on. This is what I can see:-

1. You have two boxes which joined the project about the same time, one running windoze and one running linux.
2. your windoze box has only ever downloaded one workunit which will expire in a day or two if not completed.
3. The fact that it hasn't downloaded a second workunit makes me think it's not currently running.
4. Your linux box, although slow, seems to be doing just fine. It has completed one workunit, is working on a second, and has downloaded a third, ready to start when the second finishes. It appears quite normal.

Perhaps you might like to rethink your question, particularly the bit about the three Einstein processes running. I think you have one Einstein process running under Boinc and Boinc is doing its job of properly managing the flow of work to and from that Einstein process.

As for linux being slow, yes this is a known problem. There are many threads on these boards that discuss this. I'm sure you will find them with a quick check of some recent thread titles.

Cheers,
Gary.

Darren
Darren
Joined: 18 Jan 05
Posts: 94
Credit: 69,632
RAC: 0

I know what you're asking

I know what you're asking about with the 3 einstein processes, but I have no idea what it means. When running ps -e, on a couple of my linux systems I only show 1 einstein process, but usually it shows 3. 2 of the 3 seem, in essence, to be defunct (though they are not marked defunct), and only the third process actually accumulates any time. They always have 3 sequential PID numbers, with only the last number being the one really doing any running.

Sorry, but again, I don't know why it does this, but you are correct that it does do it.

On my adjacent system, I'm looking at the ps list now and it shows:
boinc419
einstein_4.80_i686-pc-linux-gnu
einstein_4.80_i686-pc-linux-gnu
einstein_4.80_i686-pc-linux-gnu
setiathome_4.02_i686-pc-linux-gnu
hadsm3_4.04_i686-pc-linux-gnu
hadsm3um_4.04_i686-pc-linux-gnu

Why 3 einstein entries, again, no clue. I've never really paid much attention to it, as it's never seemed to cause me any problems.

forloop32
forloop32
Joined: 19 Mar 05
Posts: 7
Credit: 725,230
RAC: 0

You are right about my

Message 9352 in response to message 9350

You are right about my Windoze box - it overheats when it does so many calculations, so I have turned it back off. Time to get a new one.

As for my Linux box, looks like three to me:
daver 20201 27428 0 19:28 ? 00:00:00 einstein_4.80_i686-pc-linux-gnu @conf -f946.898 -o .Ha --startTime 755500140 --endTime 755545762 // @conf -f946.898 -o .Hb --startTime 751686285 --endTime 751727775 // -1 Fstats.Ha -2 Fstats.Hb -f 0.001 -a 0.02 -d 0.02 -s 946.900 -e 947.000 -b -o polka.out

daver 20202 20201 0 19:28 ? 00:00:00 einstein_4.80_i686-pc-linux-gnu @conf -f946.898 -o .Ha --startTime 755500140 --endTime 755545762 // @conf -f946.898 -o .Hb --startTime 751686285 --endTime 751727775 // -1 Fstats.Ha -2 Fstats.Hb -f 0.001 -a 0.02 -d 0.02 -s 946.900 -e 947.000 -b -o polka.out

daver 20203 20202 99 19:28 ? 00:53:09 einstein_4.80_i686-pc-linux-gnu @conf -f946.898 -o .Ha --startTime 755500140 --endTime 755545762 // @conf -f946.898 -o .Hb --startTime 751686285 --endTime 751727775 // -1 Fstats.Ha -2 Fstats.Hb -f 0.001 -a 0.02 -d 0.02 -s 946.900 -e 947.000 -b -o polka.out

I appreciate knowing why there are three WU's in progress. I was afraid that it was working on two or three at a time.

Thanks for your help.

> > Does all of this make sense?
>
> Unfortunately, no. I don't really understand what you are asking and I don't
> think you have a correct idea of what is going on. This is what I can see:-
>
> 1. You have two boxes which joined the project about the same time, one
> running windoze and one running linux.
> 2. your windoze box has only ever downloaded one workunit which will expire in
> a day or two if not completed.
> 3. The fact that it hasn't downloaded a second workunit makes me think it's
> not currently running.
> 4. Your linux box, although slow, seems to be doing just fine. It has
> completed one workunit, is working on a second, and has downloaded a third,
> ready to start when the second finishes. It appears quite normal.
>
> Perhaps you might like to rethink your question, particularly the bit about
> the three Einstein processes running. I think you have one Einstein process
> running under Boinc and Boinc is doing its job of properly managing the flow
> of work to and from that Einstein process.
>
> As for linux being slow, yes this is a known problem. There are many
> threads on these boards that discuss this. I'm sure you will find them with a
> quick check of some recent thread titles.
>

forloop32
forloop32
Joined: 19 Mar 05
Posts: 7
Credit: 725,230
RAC: 0

Thanks for the confirmation.

Message 9353 in response to message 9351

Thanks for the confirmation. You can see from my previous reply that all three of my processes seem to be accumulating time - 19:28. I think that I'll continue investigating. I'll let you know if I discover anything.
Thanks again.

> I know what you're asking about with the 3 einstein processes, but I have no
> idea what it means. When running ps -e, on a couple of my linux systems I
> only show 1 einstein process, but usually it shows 3. 2 of the 3 seem, in
> essence, to be defunct (though they are not marked defunct), and only the
> third process actually accumulates any time. They always have 3 sequential
> PID numbers, with only the last number being the one really doing any
> running.
>
> Sorry, but again, I don't know why it does this, but you are correct that it
> does do it.
>
> On my adjacent system, I'm looking at the ps list now and it shows:
> boinc419
> einstein_4.80_i686-pc-linux-gnu
> einstein_4.80_i686-pc-linux-gnu
> einstein_4.80_i686-pc-linux-gnu
> setiathome_4.02_i686-pc-linux-gnu
> hadsm3_4.04_i686-pc-linux-gnu
> hadsm3um_4.04_i686-pc-linux-gnu
>
> Why 3 einstein entries, again, no clue. I've never really paid much attention
> to it, as it's never seemed to cause me any problems.
>
>
>

Darren
Darren
Joined: 18 Jan 05
Posts: 94
Credit: 69,632
RAC: 0

> Thanks for the

Message 9354 in response to message 9353

> Thanks for the confirmation. You can see from my previous reply that all
> three of my processes seem to be accumulating time - 19:28. I think that I'll
> continue investigating. I'll let you know if I discover anything.
> Thanks again.

I may be looking at your columns a little out of whack, but I think only your last process is accumulating time. You have process IDs 20201, 20202 and 20203, then the six digit number (xx:xx:xx) immediately before the process name should be it's accumulated run time.

That would give you 00:00:00 for PID 20201, 00:00:00 for PID 20202, and 00:53:09 for PID 20203. Unless your columns are maybe non-standard for what I'm used to on mandrake.

Wurgl (speak^Wcrunching for Special: Off-Topic)
Wurgl (speak^Wc...
Joined: 11 Feb 05
Posts: 321
Credit: 140,550,008
RAC: 0

Guys, you have a Linux 2.4

Guys,

you have a Linux 2.4 kermel running. With a 2.6 kernel you see only boinc and einstein as a process.

The reason for this difference is called thread. And one of the big changes between the 2.4 and the 2.6 kernel is the handling of threads.

The number cruncher program, einstein, is a multi-threaded program. Just the same as mozilla where I see 4 entries in the ps list on my 2.4 kernel.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 4,888
Credit: 29,325,491,074
RAC: 35,237,105

> You are right about my

Message 9356 in response to message 9352

> You are right about my Windoze box - it overheats when it does so many
> calculations, so I have turned it back off. Time to get a new one.

Actually, that box wouldn't be too bad if you could solve the heat problem. Check your cpu heatsink and fan for accumulated dust. I've just noticed that it's listed as a mobile AMD 1500+ so I guess that means it's a laptop and likely to have heat issues anyway.

I'm also glad to see that others were able to work out why you were seeing three processes. Wurgl's reply sorted the whole issue completely.

Good luck with your crunching. Investigate running the windoze client under WINE to improve your crunching rate.

Cheers,
Gary.

gravywavy
gravywavy
Joined: 22 Jan 05
Posts: 392
Credit: 68,962
RAC: 0

THere are two issues raised

THere are two issues raised in your intitial posting, one about different WU shown as 'in progress' on the website, and the other about different processes inside your machine.

They both amount to 3 at present, which makes it look like they might be related, but this is a misleading coincidence.

The website describes a WU as 'in progress' from the time it is issued to you until it is either returned or times out (at present it times out 7 days later). The website actually has no idea whether your box is processing that WU right now, or even if your box is switched on.

On the machine itself, you will have several processes: one coordinating BOINC-wide scheduling (in case you are running multi-projects), one handling E@H-specifc scheduling, one handling communication with the servers, one handling communications with you, the user and so on.

One reason for doing this is that it is easier to write software if you split the jobs into different process, another is that this allows different processes/threads/tasks to be run at different priorities. (The number crunching will be low priority / high 'nice' numbers, but other parts of the E@H still need to run at a normal priority, like uploading/downloading work and responding to the user)

Anyway, the general idea is there are good reasons for spiltting the work into processes or threads, but that the number of processes or threads is not connected to the number of WU issued to you.

hope that helps

~~gravywavy

forloop32
forloop32
Joined: 19 Mar 05
Posts: 7
Credit: 725,230
RAC: 0

You are absolutely correct!

Message 9358 in response to message 9354

You are absolutely correct! Thanks. The 19:28 is probably when the process was started.

As you've seen in other responses, the other "processes" are probably threads!

Thanks for your help.

>
> I may be looking at your columns a little out of whack, but I think only your
> last process is accumulating time. You have process IDs 20201, 20202 and
> 20203, then the six digit number (xx:xx:xx) immediately before the process
> name should be it's accumulated run time.
>
> That would give you 00:00:00 for PID 20201, 00:00:00 for PID 20202, and
> 00:53:09 for PID 20203. Unless your columns are maybe non-standard for what
> I'm used to on mandrake.
>
>
>

forloop32
forloop32
Joined: 19 Mar 05
Posts: 7
Credit: 725,230
RAC: 0

Thanks to each of you. I

Thanks to each of you. I appreciate your help. I think that I now understand what I am seeing and know that I understand how BOINC and Einstein better now.

> I am new to BOINC. I was very surprised to see how slowly my Linux system
> seemed compared to others, so I started looking around. I noticed two key
> things that concern me.
>
> First, on the website I see that I have two results in progress. Having a
> single CPU system, I expected only one.
>
> Second, I see that there are three Einstein processes running, all with the
> same parameters. The boinc process starts the first, which starts the second,
> which starts the third.
>
> Does all of this make sense?
>
> Thanks!
>
>

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.