2nd Weekend Crunch

Post by **UBT - Timbo** » Wed Mar 15, 2006 11:25 am

Hi all,

Background:
On the Weekend of 25th-26th March, we're holding a 48 hour "Weekend Crunch", where volunteers from the team will be crunching for one chosen BOINC project.

Following the successful 1st Crunch (where we crunched solely for SETI@home), the 2nd Crunch features the Rosetta project (with Einstein chosen as a "back-up" in case Rosetta runs out of work or the site is down).

I've been crunching Rosetta for the last day or so to see if there are going to be any issues for our forthcoming "Crunch".

Work seems to be available, but it does seem a bit sporadic. I'm sometimes getting "No work from project" as well as "Couldn't connect to hostname (boinc.bakerlab.org)".

But hopefully, in around 10 days time, we should be OK for work.

Of course, you could fill your cache beforehand....!

Secondly, due to a problem with Outlook on my PC, I've not been able to send out the invitation to all the team members advising them of the chosen project for the 2nd Crunch. Hopefully will have this done later today, so keep an eye on your Inbox for the "Crunch announcement" and if you feel the need, please forward to any of your friends who have computers to get Crunching...!

regards,

Tim

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Wed Mar 15, 2006 12:54 pm

Dazza has put it on our shiny calender also

Post by **UBT - Mikee** » Wed Mar 15, 2006 9:55 pm

Well I'd love to participate but I won't be able.
Since early this year EVERY Rosetta I've downloaded has either:-

1 - 'Unrecoverable error' if I use the computer when Rosetta is in the middle of a run and I come to use the computer or

2 - I get a "1% continual loop"

I suspect both faults are caused by the same bug which I believe they are trying to hunt down. Just tried 4.82 with the same results - Ho Hum...

I'll just sit on the sidelines and watch.

Can I do Einstein instead please?

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Wed Mar 15, 2006 9:58 pm

mikeejones1 wrote:Can I do Einstein instead please?

If your on windows make sure you have the optimised app see the thread about this in the forum.

Rom from SETI@Home is over at Rosetta as a consultant trying to fix the rest of there errors for them so hopefully soon those things will be blasted to the past, if anyone can fix it ROM can

Post by **UBT - Timbo** » Wed Mar 15, 2006 11:04 pm

mikeejones1 wrote:Can I do Einstein instead please?

Mike,

Please feel free to do whichever project you feel capable of returning credits on...if Rosetta causes you an issue (and I've returned some that failed as well!) then please feel free to maybe work on Einstein.

As mentioned before, the whole point of doing a "Crunch" over a weekend is to generate a little bit of Team spirit, so that members feel part of a "band of crunchers" all doing the same project (at the same time) to benefit the project chosen.

Ultimately, though, ANY credits gained during the Crunch period wil count, irrespective of which project you feel happiest at crunching for...!

regards

Tim

PS Welcome to the forum!

Post by **UBT - Timbo** » Thu Mar 16, 2006 10:33 am

Instructions if you do want to take part:
-----------------------------------------
1) Sometime (on/before 24th March), is to make sure you are already a member of the Rosetta project - visit our website and click on the "Join our Rosetta team here" link from our website front page.

If you are not a member of Rosetta, then you will need to join up to the project first. You can do this within BOINC Manager, by "pasting" the following URL

http://boinc.bakerlab.org

into the first box when you click on "Attach to project"

Then specify your "normal" email address that you used when you signed up to other BOINC projects and a memorable password. BOINC Manager will then download the relevant application for you.

Or you can click here: http://boinc.bakerlab.org/rosetta/creat ... t_form.php

Once, you've joined to the project, you then need to join our team.

Do this here:

http://boinc.bakerlab.org/rosetta/team_ ... php?id=151

2) Then set all your other projects to "No new work" as well as "suspending" them from actually doing any crunching during the week-end.

This will prevent you downloading work, which may reach it's deadline BEFORE you've had a chance to complete the Rosetta work units.

3) Set the Rosetta project as the only one able to download new work.

This will then allow BOINC to "concentrate" on the work that Rosetta has available. Note that sometimes, projects may run out of work to download, so you may need to check your PC to ensure you have enough work. (You can check this by clicking on the "Work" tab in BOINC and check to see how much work you have. If you run out, simply re-enable one or more of your other projects - there's no point spinning your wheels waiting for something to download, if you have no work at all.

4) On/after 27th Feb, you can switch each of the other projects back to "Allow new work" and "Resume".

Once the weekend is over, then you simply wait for Rosetta to finish returning all of the work it's completed, and then re-enable your other projects. You may even want to continue crunching for Rosetta after the Crunch finishes.

If any questions, please join the forum (to post a question for any of our team members to answer) or email me back at the address I used to send you the invitation.

regards and Happy Crunching

Tim
Founder - UK BOINC Team

DJH@GB-Ro · Post by **DJH@GB-Ro** » Thu Mar 16, 2006 4:30 pm

mikeejones1 wrote:2 - I get a "1% continual loop"

I got that twice today, it hasn't happened before. Is it likely to keep happening?

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Thu Mar 16, 2006 4:45 pm

DJHunter wrote:
mikeejones1 wrote:2 - I get a "1% continual loop"
I got that twice today, it hasn't happened before. Is it likely to keep happening?

It's a known problem with some WU's, try opening the graphic for a few mins then shut off the graphics this can some times help.

It will keep happening until the error can be found

DJH@GB-Ro · Post by **DJH@GB-Ro** » Thu Mar 16, 2006 4:58 pm

I'm stuck on 1% for my 3rd WU of the day

I've opened the graphics and will shut them off in a few minutes

Actually, I think one was just a error, not eh 1% thing

Post by **UBT - Timbo** » Thu Mar 16, 2006 5:53 pm

UBT - Halifax--lad wrote:Its a known problem with some WU's, try opening the graphic for a few mins then shut off the graphics this can some times help.

It will keep happening until the error can be found

Hi David,

I get this problem as well - except for me BOINC is set up as a service and so it doesn't show the graphics at all (for any project !!). So, I can't use your "trick".

Can I change over to run this as a "user" easily, or is it "RE-INSTALLATION" time...!

regards,

Tim

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Thu Mar 16, 2006 6:44 pm

I think you have to re-install BOINC to change it to single user mode.

Hopefully very shortly ROM will find the problems, there must be something wrong some where with that bloody code

DJH@GB-Ro · Post by **DJH@GB-Ro** » Thu Mar 16, 2006 6:48 pm

I keep getting the 1% thing. What should I do with these WUs? Just abort them or what?

Post by **UBT - Timbo** » Thu Mar 16, 2006 6:49 pm

UBT - Halifax--lad wrote:I think you have to re-install BOINC to change it to single user mode.

Hopefully very shortly ROM will find the problems, there must be something wrong some where with that bloody code

OK - Not a problem for me - done it many times before...!

Do hope they sort out Rosetta soon (assuming it's a BOINC/Rosetta software clash).

If not, then we might have to change to Einstein for the forthcoming "Crunch".

regards,

Tim

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Thu Mar 16, 2006 6:57 pm

Another suggestion by Rosetta is to increase the CPU time option to something higher than what you have it set already, the problem seems to appear only when the CPU run time is low, I have no problems on 8 hrs of CPU run time

Darren · Post by **Darren** » Thu Mar 16, 2006 7:28 pm

Sometimes mine take about 30 mins to move off the 1% but for one it got stuck and the time was not moving either! so it would not matter how long I had the max run time to be.

Mark L. · Post by **Mark L.** » Thu Mar 16, 2006 10:13 pm

Just a thought Master...

If Rosetta is known to have an ongoing problem at this time, which is being worked on, wouldn't it make sense to run the backup (Einstein) this time anyway, and make Rosetta the weekend crunch subject for next month?

Otherwise you may just have lots of people trying to crunch faulty units and our crunch level may actually go DOWN because people stopped their other projects for the weekend?

I'm only suggesting it now, because:-

1. You still have time to get the word out

2. Einstein seems to be pretty stable at present and also has an optimised app for those who want to risk it!

Just my thoughts

Mark L.

Post by **UBT - Timbo** » Thu Mar 16, 2006 10:27 pm

UBT - Mark L. wrote:Just a thought Master...(snip) Mark L.

Perfectly well thought out.

I've already sent out a "blanket" email to every team member who hadn't already "unsubscribed" that the 2nd Crunch would be Rosetta.

And I have tried before and then this time to give people at least 10-14 days notice.

This way, as you rightly thought, there's time to figure what's going on and if people are going to have problems.

I'm already "testing the ground" by crunching Rosetta and keeping an eye out for "issues".

I'll probably give it one or two more days, and if problems still seem to be occuring (on my WU's and possibly others) then we might need to change tact beforehand, so as to not "lose" two days worth of crunching to a "flaky" project.

We'll see if anyone else is having problems with Rosetta and we'll see what transpires over the next couple of days.

My guess is that IF we need to "change" then we need to notify peoiple by end of this weekend...!

regards

Tim

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Sat Mar 18, 2006 1:52 pm

From Rosetta:

On the error fixing front there has been rapid progress! I hope to have very good news in the next few days.

Post by **UBT - Timbo** » Sun Mar 19, 2006 10:37 am

UBT - Halifax--lad wrote:From Rosetta:On the error fixing front there has been rapid progress! I hope to have very good news in the next few days.

Hi David,

I've seen the Rosetta project forum and know that Rom is working on a fix - I undertstand it's something to do with the way that the Rosetta "app" closes when BOINC tells it to switch to a different project - my observations have also recoginised this, as I've seen this problem once or twice.

But the question is: In 5 and a bit days time, we've asked our team members to run a project that is a bit flaky right now.

For those who have NO PROBLEMS (like yourself) then there's no issue.

But those who either are having problems now (like me and some others) or who might be "newbies" who join up for the crunch, suspend their other work, and then find NO ROSETTA WORK IS SUCCESSFULLY RETURNED and they only get problems, then this is a BIG ISSUE.

So, I'm thinking that I need to send out an email TODAY to warn people about this potential problem.

I'm not keen on abandoning the Crunch - but I'm getting to the point where I think we need to take a clear step and recommend that members crunch for the BACK-UP project, Einstein, if Rosetta is causing them issues.

So I hope that some members will continue on Rosetta, if all's well with them.

But I am concerned that we could be making problems for people and hence we need to have a plan of action, which we can implement soonest, in order to save people "spinning their wheels".

I will put together a draft and "chew on it" over lunch.

regards,

Tim

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Sun Mar 19, 2006 10:44 am

Here is another update, the next Roestta App is going to report back the correct files it needs in order to find any remaining errors

For some exciting news on the error front, and a description of the problem and the solution, take a look at
http://www.romwnet.org/dasblogce/

It is amazing how fast Rom brought the error rate way down on Windows machines on Ralph, and we have spent today and yesterday making further modifications based on his suggestions that will hopefully bring the error rate down still further. You should see the improvements made thus far on rosetta@home by Tuesday.

Post by **UBT - Timbo** » Sun Mar 19, 2006 11:17 am

UBT - Halifax--lad wrote:For some exciting news on the error front, and a description of the problem and the solution, take a look at
http://www.romwnet.org/dasblogce/

Hi David,

Hadn't seen that one - good on yer !!

OK - So we might be OK for next weekend....?

But we won't know maybe until Tuesday......!

(And reading through the blog, seems Einstein might also be affected in a similar way - though not as big as Rosetta is now !!)

Typical !

OK - So, we'll just have to wait until Tuesday and see - happy to postpone any email announcements until then and we'll take a view.

Good that it's being fixed...!

regards,

Tim

Post by **UBT - Mikee** » Sun Mar 19, 2006 11:26 am

Maybe a good time for a qucik 2 day poll on Einstein then?

Darren · Post by **Darren** » Tue Mar 21, 2006 4:08 pm

If we are going to be comparing weekend#1 with weekend#2 then, to keep things fair, we should start at 11:00pm on Friday.
Don't forget the clocks go forward an hour at 1:00 on Sunday!

Post by **UBT - Timbo** » Tue Mar 21, 2006 5:08 pm

Dazza wrote:Sometimes mine take about 30 mins to move off the 1% but for one it got stuck and the time was not moving either! so it would not matter how long I had the max run time to be.

Hi all,

If you have "stuck" Rosetta WU's keep an eye on this thread:

http://boinc.bakerlab.org/rosetta/forum ... php?id=899

Basically, IF you want, ONLY run RALPH WU's for now instead of Rosetta WU's......!

Then Rom (Walton) and the admins can see if they can fix things.

regards,

Tim

UBT - BHCJackie · Post by **UBT - BHCJackie** » Tue Mar 21, 2006 6:23 pm

UBT - Timbo wrote:If you have "stuck" Rosetta WU's keep an eye on this thread:

http://boinc.bakerlab.org/rosetta/forum ... php?id=899

Ok, I'm totally confused by the thread Tim mentioned.
AIUI the RALPH people want users to change "Leave applications in memory while preempted?" to "Yes". But if you do this won't it affect the other projects one is attached to?
I honestly don't understand about suspended applications consuming swap space but if the projects were meant to run this way wouldn't the default be "Yes" rather than "No"?

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Tue Mar 21, 2006 6:29 pm

There isn't a default it is a user option of either yes or no, BOINC doesn't need to leave anything in memory except for Rosetta & uFluids at the moments as they don't work correctly

If you have it set to No and don't run these projects then it makes little difference to the WU's

All it means if it was set to Yes is WU's will start off again from where they finished rather than the last checkpoint if set to No

UBT - BHCJackie · Post by **UBT - BHCJackie** » Tue Mar 21, 2006 6:44 pm

UBT - Halifax--lad wrote:There isn't a default it is a user option of either yes or no

I realise that mate, but any total noob joining any project just sees it as a default. It only becomes a user option if you click on edit preferences and many people won't.

TBH I always click on edit links simply to see what I can and can't change but other people, who aren't as nosy as I am, leave things like that well alone so they don't "break" anything.

So what exactly does consuming swap memory do? If I change to yes will GertII grind to a halt or what?

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Tue Mar 21, 2006 6:48 pm

Keeping in memory does exactly what it says on the tin all the WU info will me left in memory so when BOINC switches to another WU, the previous WU is held in memory, this ensures the WU starts off again from exactly the same place it finished at when the WU switch happened.

If set to No you can lose valuable time as the project doesn't stay in memory and will only start from its last checkpoint that was written.

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Tue Mar 21, 2006 7:16 pm

Ok so was in the bath and thought I had better put a simple to follow example

Preempt in Memory = No

If a WU checkpoints every 10 mins and then switches to another WU at 29 mins in then when it comes back to this WU again it would start from 20 mins

Preempt in Memory = Yes

If a WU checkpoints every 10 mins and then switches to another WU at 29 mins in then when it comes back to this WU again it would start from 29 mins

If you have 512Mb of memory your computer will happily run most projects and keep in memory, so long as you don't have something running that uses loads of memory.

Mine did have this amount in and crunched 22 projects happily

UBT - BHCJackie · Post by **UBT - BHCJackie** » Tue Mar 21, 2006 8:25 pm

UBT - Halifax--lad wrote:Ok so was in the bath and thought

Did you jump up, splosh water all over the floor and shout "Eureka!"?

I hope you didn't go runing down the street dressed only in your bath towel :sign5:

UBT - Halifax--lad wrote:
If you have 512Mb of memory your computer will happily run most projects and keep in memory, so long as you don't have something running that uses loads of memory.

Mine did have this amount in and crunched 22 projects happily

Ok, I get it now. TY David.
I'll change my preferences to "Yes".

Darren · Post by **Darren** » Wed Mar 22, 2006 7:03 pm

So, now that I have a huge cache of Rosetta work units. Which project are we crunching for, as a team, this weekend?

UBT - BHCJackie · Post by **UBT - BHCJackie** » Thu Mar 23, 2006 12:31 am

I wrote:I'll change my preferences to "Yes".

So I did, crunched a couple more WUs and I'm still getting errors.

Sorry Rosetta but I'm not playing anymore

I've suspended Rosetta and it will remain that way until there is a fix.

UBT - Halifax-lad · Post by **UBT - Halifax-lad** » Thu Mar 23, 2006 7:38 pm

I'm running Ralph instead this weekend should get a few done for them on the new client, got the latest version of BOINC so at least any error reports will be sent back to them correctly

Post by **UBT - Timbo** » Thu Mar 23, 2006 10:34 pm

UBT - Halifax--lad wrote:I'm running Ralph instead this weekend should get a few done for them on the new client, got the latest version of BOINC so at least any error reports will be sent back to them correctly

Hi David,

I'm already running Ralph, as had lots of problems with Rosetta in the last week (was testing that "all was well", which it wasn't !!)

Have latest "development" version of BOINC (v5.3.28 - as recommended by Rom) and now have some 4.94 WU's.

So far, no problems but I still get a weird 1% progress as soon as the WU starts - only changes after a while - so will be interesting to see what happens once the WU's are returned.

Have now got s39l opti client for Einstein and will start crunching those WU's already in the cache of this PC from tomorrow. Been crunching Einsteins on another couple of PC's for the last few days in order to build up a "head of steam" on Einstein...!

Now here's a strange one for you to think about:

I have "A36 opti" on a P3 running at 600MHz - works well for all other projects, but currently it's working on a single Einstein WU - it's at 83% after, wait for it, ...........77 hours !! Still 13 hours to go (it says). Hope I get some credit for that one....!

regards,

Tim

UK BOINC Team Forum

2nd Weekend Crunch

2nd Weekend Crunch

2nd Weekend Crunch - instructions