1) Message boards : Number crunching : 362,000 secs per 1,000,000 turns (Message 12727)
Posted 13 Feb 2006 by Profile Purple Rabbit
Post:
I have never seen a 286, 386 or 486 chip crunching except for one 486 working on SETI but it is taking close to a week per WU there. Definately not good for projects with shorter deadlines.

My 486 (AMD 586-133) does SETI in 12.5 days. It's not really worth it. SIMAP seems to be a good choice for slower computers. The 486 does it in 22 hours and I'm almost always first to return a result.
2) Message boards : Number crunching : Database Errors (Message 12065)
Posted 15 Jan 2006 by Profile Purple Rabbit
Post:
It is nearly impossible to come up with a "rule" that will clearly identify a "bad" entry from a system perspective.

That's true Paul, but I think a floating point value in a field where it should say "Linux" is a prime candidate :-)

If the computer has no results on it, you can delete it yourself.

Unfortunately the computer still has one result from August that will probably never go away. It's complete, but hasn't been deleted.

I didn't realize that this was really hard for the admin to fix. As Chrulle said it's not affecting the project. I agree what I have is just cosmetic. I reported the error and life goes on :-)
3) Message boards : Number crunching : Database Errors (Message 11984)
Posted 14 Jan 2006 by Profile Purple Rabbit
Post:
Well, the spurious database entries are not affecting the functioning of the system so it is not a big priority to fix it.


OK, I now know your database philosophy. No big deal :-)

I wasn't sure if you knew about the errors. These things are annoying to the users (at least me) tho :-) At least please put it on your "to do" list. I was afraid this might be part of a larger problem.
4) Message boards : Number crunching : Database Errors (Message 11978)
Posted 14 Jan 2006 by Profile Purple Rabbit
Post:
User-credit is separate from host-credit, so it's no problem to delete any old hosts there all results has been purged.


Yes, that's all I ask! Delete the darn thing and let the world go on :-) Again, I think LHC needs to look at the datbase.

To get my message back to what I intended: The database crash in August caused some problems. I think a rebuild might be in all of our best interests. There are still some artifacts left over from the crash.

I'm seeing some fallout in my accounts, but I suspect there's more. I had the unfortunate timing to change my Linux OS at the same time LHC was having problems...sigh.

Otherwise I'm doing fine as a cruncher. The world goes on.
5) Message boards : Number crunching : Database Errors (Message 11974)
Posted 14 Jan 2006 by Profile Purple Rabbit
Post:

4) They do need to expunge such entries, however doing so would alter the total amounts of credits have been issued once those hosts have been removed (BadThing™).

Travis, again thank you. My database entry for CPU type for 33305 is:

1.270598
0
My entry for Operating system is:

0
1458205218.5383

These fields are clearly corrupted in the database when you might expect ascii test in these fields. Thank you for your response, but I think this is a sysadmin problem.
6) Message boards : Number crunching : Database Errors (Message 11972)
Posted 14 Jan 2006 by Profile Purple Rabbit
Post:
In another post, long, long ago....


Travis, thank you, but this doesn't apply to what I was saying. My host (33305) was active in August. It is now host #68704 and doing well. Unfortunately I updated Tomato (33305) to a new Linux distribution and killed all the BOINC files just when the LHC database crashed. I started over, but the old, corrupted file remained in the database. It can't be changed because Tomato has a new identity.

It also looks like some other files pre-August also remain. It looks like the database is corrupted to me.

I say again, there are extraneous entries in the database. An LHC sysadmin should look at this because I assume the impact is larger than what I can see.
7) Message boards : Number crunching : Database Errors (Message 11968)
Posted 14 Jan 2006 by Profile Purple Rabbit
Post:
I think LHC may want to do a database search for invalid and out of date entires. I have one host (33305) that has corrupted CPU type and Operating system entries. This host now has a new identity, but I can't merge the hosts because of the database error.

These are from last August when the database went down. There are also some strangler WU that either are pending or or have been granted credit, but not deleted or acknowledged. Most are from before August.

Maybe a database scrub is in order? There seems to be some leftovers in here.
8) Message boards : Number crunching : New WUs (Message 6884)
Posted 8 Apr 2005 by Profile Purple Rabbit
Post:
My Windows machines got version 4.66. The Linux machine got version 4.63. I returned one on the Linux machine for 0 claimed credit. I got 6.6. Thank you Windows guys :-)



©2024 CERN