[Shaktai]

Predictor is back up and starting to trickle out some work units. There are still a number of "download failures". You may need to kick your boxes to get them to download. If you get some failures then just kick them again. (Figuratively that is. I am not liable for any broken feet, stubbed toes or damaged computers.)

Let's get crunching. There are teams to pass.

[buggsuperstar]

I'm getting this message from my BIONIC client:

5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Message from server: No work sent
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Message from server: (there was work but it was committed to other platforms)
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Message from server: (your core client is out of date - please upgrade)
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|No work from project
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Deferring communication with project for 1 days, 0 hours, 0 minutes, and 0 seconds

Is this normal?

This is my first time using BIONIC by the way, so pardon the (potentially) silly question.

I seem to be crunching SETI just fine.

[Shaktai]

Originally Posted by buggsuperstar

I'm getting this message from my BIONIC client:

5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Message from server: No work sent
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Message from server: (there was work but it was committed to other platforms)
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Message from server: (your core client is out of date - please upgrade)
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|No work from project
5/25/2005 6:53:08 AM|ProteinPredictorAtHome|Deferring communication with project for 1 days, 0 hours, 0 minutes, and 0 seconds

Is this normal?

This is my first time using BIONIC by the way, so pardon the (potentially) silly question.

I seem to be crunching SETI just fine.

Not a silly question. The BOINC server is still having some issues. Predictor uses a different validation process then other projects. Windows work is validated only agains other windows work, Linux against Linux and Mac against Mac. To make it more complicated, they also associate by CPU type, so AMD only against other AMD and Intel against Intel, etc. As a result, occasionally they will run out of work for a particular platform. It is a short term issue normally.

The second issue of (your core client is out of date - please upgrade) is a fluke in itself. The message is not correct, but is caused by a server mismatch of some kind. Just ignore it. Predictor has typically been a pretty stable project, but this past weekend they had some problems and it will probably take a few days to get all the issues resolved. In the mean time, if you don't have work from predictor, then BOINC will keep crunching SETI and your machine will stay busy. Once you get work for both projects, then it will alternate between the two.

TIP. With predictor, keep your cache below 3 days maximum if running two projects. (presuming that you crunch 24 hours a day) The deadline is only 7 days compared to 14 days for SETI. If you download too much work, you will never finish it in time to get credit.

[buggsuperstar]

Originally Posted by Shaktai

Not a silly question. The BOINC server is still having some issues. Predictor uses a different validation process then other projects. Windows work is validated only agains other windows work, Linux against Linux and Mac against Mac. To make it more complicated, they also associate by CPU type, so AMD only against other AMD and Intel against Intel, etc. As a result, occasionally they will run out of work for a particular platform. It is a short term issue normally.

The second issue of (your core client is out of date - please upgrade) is a fluke in itself. The message is not correct, but is caused by a server mismatch of some kind. Just ignore it. Predictor has typically been a pretty stable project, but this past weekend they had some problems and it will probably take a few days to get all the issues resolved. In the mean time, if you don't have work from predictor, then BOINC will keep crunching SETI and your machine will stay busy. Once you get work for both projects, then it will alternate between the two.

TIP. With predictor, keep your cache below 3 days maximum if running two projects. (presuming that you crunch 24 hours a day) The deadline is only 7 days compared to 14 days for SETI. If you download too much work, you will never finish it in time to get credit.

Thanks Shaktai.

My BIONIC client finally managed to get a Predictor workunit and it happily crunching it as we speak.

I'm devoting the majority of my resources to Predictor since Team MacNN is currently holding court at the top spot. I'd like to help keep it that way.

GO TEAM!

[mikkyo]

Good thing Predictor is sending out workunits again as our new member needs many to crunch.

[Shaktai]

Originally Posted by mikkyo

Good thing Predictor is sending out workunits again as our new member needs many to crunch.

Yep, it looks like it is almost back to its normal self, and I am back to getting kicked farther down the list. (Just can't get any respect)

Hey buggssuperstar, hope you're still with us. The work is there, the clients are back to normal and the servers are running well. Looks like some of the PC crunchers bailed out for a while, and now is a great time for our push. We're currently outcrunching everyone by a good margin.

[buggsuperstar]

Originally Posted by Shaktai

Yep, it looks like it is almost back to its normal self, and I am back to getting kicked farther down the list. (Just can't get any respect)

Hey buggssuperstar, hope you're still with us. The work is there, the clients are back to normal and the servers are running well. Looks like some of the PC crunchers bailed out for a while, and now is a great time for our push. We're currently outcrunching everyone by a good margin.

Hey Shaktai,

Thanks for the heads up. I had switched over to SETI full time since I was getting no work from Predictor. I've since switched back to Predictor full on.

Let's put more distance between us and the competition.

[joe_it]

Originally Posted by buggsuperstar

Hey Shaktai,

Thanks for the heads up. I had switched over to SETI full time since I was getting no work from Predictor. I've since switched back to Predictor full on.

Let's put more distance between us and the competition.

I am glad you guys are back to normal. I am still having issues. I have a few *cough cough* windows clients that I use and they have been barfing on me lately. They will get to xx% and then just stop. Just this morning I had a work unit that was at 47% and it just stopped right there. It had been running for a few days at 47% and since it is a machine I don't frequent much, I was a bit upset. I am running 4.43 of the Boinc Manager on that computer. I tried downgrading to 4.19 on another machine with the same problem, no luck. It is really making me mad here..

Anyone else having similar issues and or know of a fix for me?

Joe

[joe_it]

Oh ya, another thing. ON my G5, I am running boinc menubar 4.43. I have my account set-up to download half a days worth of work. I REPEATEDLY get a message saying not enough work, downloading more work, and it just scrolls across the log screen, NON-STOP. Have others with G5's had issues?

Joe

[buggsuperstar]

Originally Posted by joe_it

I am glad you guys are back to normal. I am still having issues. I have a few *cough cough* windows clients that I use and they have been barfing on me lately. They will get to xx% and then just stop. Just this morning I had a work unit that was at 47% and it just stopped right there. It had been running for a few days at 47% and since it is a machine I don't frequent much, I was a bit upset. I am running 4.43 of the Boinc Manager on that computer. I tried downgrading to 4.19 on another machine with the same problem, no luck. It is really making me mad here..

Anyone else having similar issues and or know of a fix for me?

Joe

Well, I have a similar issue with my system which is also a Windows box.

What happens is when switching between work units, it will fail to resume crunching the first unit after it has completed the second. An illustration:

• Work Unit 1 crunching, 70% complete
• Client stops crunching Work Unit 1 and begins crunching Work Unit 2, which has an earlier deadline
• Work Unit 2 completes crunching
• Client resumes Work Unit 1, but 1 hour later it's still at 70%. Not moved an inch.

In this case, I simply junk the unit and start crunching a new one.

I think it has something to do with this:
5/31/2005 10:01:28 PM|ProteinPredictorAtHome|Pausing result h0009A_1_93815_1 (removed from memory)

Once it is removed from memory, it is then unable to resume when it comes back to it.

Wonder if this is a bug.

<edit> forgot to add that my client is 4.43

[Shaktai]

Originally Posted by buggsuperstar

Well, I have a similar issue with my system which is also a Windows box.
...snip.....

I think it has something to do with this:
5/31/2005 10:01:28 PM|ProteinPredictorAtHome|Pausing result h0009A_1_93815_1 (removed from memory)

Once it is removed from memory, it is then unable to resume when it comes back to it.

Wonder if this is a bug.

<edit> forgot to add that my client is 4.43

Yep this is a known bug with mfold 4.33. A lot of discussion with it on the predictor forum.

Two things you can do. Make sure you keep Predictor in memor on windows until you complete your mfold 4.33 work or abort any work units using mfold 4.33. They have reverted to mfold 4.28 because of this problem. When you abort the 4.33 work, it will download the mfold 4.28 app. (4.29 had a graphics issue). That will eleminate that problem until they find the cure.

With the 4.33 the only two things that seemed to help was to "not switch between projects" or to quit and restart, or make certain the app stays in memory when paused (general preferences).

At this point, I would say check your queue. Under the "Work" tab look in the column that says "Application" Widen it out until you can see the full name. If it says mfold 4.32 or 4.33, abort it (using the button on the left side if you are using the latest boinc version.) Then it will download 4.28 and compatible work.

Because the problem was with the mfold application, not the BOINC core client, that is why reverting to an earlier client wouldn't fix the problem.

[Shaktai]

Originally Posted by joe_it

Oh ya, another thing. ON my G5, I am running boinc menubar 4.43. I have my account set-up to download half a days worth of work. I REPEATEDLY get a message saying not enough work, downloading more work, and it just scrolls across the log screen, NON-STOP. Have others with G5's had issues?

Joe

I had played with the BOINC menubar client and didn't care for it. It was a little flakey. Instead I use the full Advanced GUI and it works fine. It also provides most of the menubar functionality with additional capabilities and stability. Just close BOINC manager when you don't need it to get the best performance. Also, the predictor graphics and screensaver do not work with the Mac version. 4.43 of the advanced GUI is the recommended version along with 4.43 of the CLI. BOINC menubar still seems to be "development phase.

[buggsuperstar]

Originally Posted by Shaktai

Yep this is a known bug with mfold 4.33. A lot of discussion with it on the predictor forum.

*snip*

Because the problem was with the mfold application, not the BOINC core client, that is why reverting to an earlier client wouldn't fix the problem.

I junked all my mfold 4.33 units and the client downloaded a whole bunch of 4.28 units, so it's all good.