Proth tests completed

[PY 222]

Guys, my apologies if this has been asked before but I have tried looking for information on this but alot of the information are rather old.

Three questions with regards to the Proth tests completed section.

1. I've been running my test for about 4 days now and it hasn't even completed yet. What is the average for a test?

2. How long will we be given to finish off a test?

3. We are currently return some kind of blocks back to the server. If after we have returned about 50% of the blocks but did not finish the test, will the blocks be discarded and the test given out to a different server?

[PCZ]

PY

Look in the SOB directory and view the sob.log file.

You will see how many blocks there are in the test and how long it is taking to do each block.
Do the math on some of your PC's to get an idea of how long they will take.

[Bok]

Depends on the PC running the test. 3-4 days sounds about right though for newish intel.

Time given to complete a test is basically infinite, as long as a block is reported at least once every 30 days. I guess it's actually (30*num blocks in test).

After 30 days of no reporting on the test, it is discarded by the server and placed in a 'dropped' tests queue for someone else to pick up. You eventually lose any score you got for that test as it has to be restarted from the beginning.

Bok

[LAURENU2]

Ya same here now for 5.5 days and only at 78.8% complete and have returned 5200 blocks Loonnnggg W/U
I had to restart one PC and it started back up from block 1 and gave the other 1/2 to PCZ to finish:cry:

[PY 222]

Ok thanks guys.

So it'll be a waste of my CPU cycles if I didn't finish the test. :umm:

Ok, I'll make a note of that. Thanks again.

[Bok]

Are your machines ghosted on a semi-regular basis which would cause this problem ??

Any news on deployment... :|party|:

Bok

[PY 222]

Are your machines ghosted on a semi-regular basis which would cause this problem ??

Any news on deployment... :|party|:

Bok

No, my machines are all just running DC apps.

I am almost ready to go with my script. Will try it out on one rack of servers first.

What I am worried about is that in the Linux environment, there is no way to tell the server to stop after it is done with the test, unlike the Windows one.

Any ideas on that?

[Bok]

I seem to recall seeing a few wrappers for it. Wouldn't be too hard I think to write a quick script which checked the .log periodically and killed it when it fetched a new test. I can work on something for you if you like...

Bok

p.s. One rack.... sometimes you :scared: me!!!

[PCZ]

Are there any monitoring apps for this ?

Something like FadSpy.
Seems to be enough info in the log files.

[Bok]

Not sure if it's been updated recently but DCMonitor (http://devel.bluetentacle.co.uk/dcmonitor/v1/) appears to do it..

Where is MerePeer!!!! I'm sure he could knock something up :)

Bok

[riptide]

Ya same here now for 5.5 days and only at 78.8% complete and have returned 5200 blocks Loonnnggg W/U
I had to restart one PC and it started back up from block 1 and gave the other 1/2 to PCZ to finish:cry:
Lauren! Is your pharm AMDs? I seem to recall your passion for X2's before. AMD's may take a little longer than their Intel cousins. Don't worry about it. Keep crunching. :)

EDIT: Are the servers for receiving work down? I can't report and our respective teams last update was kinda poor!

[PY 222]

No chance of deploying at the moment.

Firstly, I have this libstdc++.so.5 issue to contend with.

Secondly, the server seems down at the moment and I cannot get any work.

Oh well.

[riptide]

Servers back! Lets see some action PY222 :thumbs:

[PCZ]

I tried DCmonitor, no luck though.
Progress doesn't update and worse it crashes the clients.

Seems it doesn't like SOB v2.5 much.

I think it puts a lock on the sob.log file when it updates and if the client tries to write to it, it can't and crashes.

[LAURENU2]

Lauren! Is your pharm AMDs? I seem to recall your passion for X2's before. AMD's may take a little longer than their Intel cousins. Don't worry about it. Keep crunching. :)

EDIT: Are the servers for receiving work down? I can't report and our respective teams last update was kinda poor!

YA all AMD Not worried about long runners But I am a little peeved:swear: that the W/U will not restart where it left off And getting any credit taken away if the hole W/U is not completed SUCKS.
I will be removing SOB from a few computers that can not go non stop for 10 days or better as it seems a waste to run it on them Back to DPAD for them

[Bok]

It should not restart from the beginning if stopped, though I think I remember some problem where that occasionally happens, normally you only lose a few blocks..

Bok

[PCZ]

Lauren

You should be able to stop and start the client without loosing work.
The client writes to the log file on the completion of each block so at most you should loose the block it is currently working on.



edit:
Apparently there is/was a bug which can cause you to loose work.
I have been looking at posts mentioning it but they are fairly old, maybe it's been fixed already.
If not it could explain what you observed.

[LAURENU2]

Lauren

You should be able to stop and start the client without loosing work.
The client writes to the log file on the completion of each block so at most you should loose the block it is currently working on.



edit:
Apparently there is/was a bug which can cause you to loose work.
I have been looking at posts mentioning it but they are fairly old, maybe it's been fixed already.
If not it could explain what you observed.
I am only Taking it off on about 4 nodes The ones that are used by people here just in case they lock it up or restart them
The others would only be affected by a power outage

Yes for the most part the client restarts where it left off . My PC locked and rebooted thats when I noticed that SOB started all over:cry:

[riptide]

Yes for the most part the client restarts where it left off . My PC locked and rebooted thats when I noticed that SOB started all over:cry:Hmmmmk Are you sure it was the same Block that started over? Maybe it dropped the original and downloaded another NEW block! I think one of mine may hay done that in the early stages!

[LAURENU2]

Im not sure what block go's where But it was 78% completed with the Proth test.
After reboot it started up at 0% :swear:

[NEBOJSA]

LAURENU2 check this thread out.

http://www.free-dc.org/forum/showthread.php?t=7434

[LAURENU2]

LAURENU2 check this thread out.

http://www.free-dc.org/forum/showthread.php?t=7434
Thank you that was of grate help THANKS, I will look into it tomorrow to see if I can recover it:thumbs:
But as I said the 4 PC's that are used as Human interfaces will stay on more reliable projects
I am pumping out some good numbers 4PC's will not heart our effort that much

:thumbs: Go Team Go:thumbs:
59.247T yesterday just 4 PC's short of making 60.00T :slap:

[Bok]

:woot:

And I can get to my 2 remote P4 3.2's tomorrow finally and get them straightened out..

Bok

[IronBits]

Incredible mass of computing power being directed at one project! :rock:

[CaptainMooseInc]

I remember when I got my Dual Xeon rig off the ground that I wanted to run 4 instances of SoB and every time the client got restarted it dropped the old test and would download a new one. Hence the fact I'm not on SoB and on DNET OGR25 now.

-Jeff

[LAURENU2]

Incredible mass of computing power being directed at one project! :rock:
And all by 20 or so members We are pumping out 2x's + the points from teams with 3 X's our active members

Yes a Incredible mass of computing power being produced

[LAURENU2]

yippy:banana: I completed my first test 100%:banana:

[Bok]

That's amazing really, that you are almost at 50T and only 1 test complete...

just shows how much power you have put on it!!!!

Way to go!

Bok

[PY 222]

yippy:banana: I completed my first test 100%:banana:

Congrats Lauren.

Hopefully this is the first of many many many more to come. :thumbs:

[PCZ]

Congrats on completing your first test :thumbs: