Bug reports for 5.84

Message boards : RALPH@home bug list : Bug reports for 5.84

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Rhiju
Volunteer moderator
Project developer
Project scientist

Send message
Joined: 14 Feb 06
Posts: 161
Credit: 3,725
RAC: 0
Message 3506 - Posted: 6 Dec 2007, 18:07:37 UTC - in response to Message 3504.  

We're looking into it -- I just contacted Rob, who can probably give you an update on these tests.

5 x the same error as BigMike mentioned ("Incorrect function. (0x1) - exit code 1 (0x1), ERROR:: Exit from: .fragments.cc line: 465"): results 689005, 688961, 688506, 688423, 689759, named "1****_BOINC_ABINITIO_BEST25_VF_SCORE3-1*--1****-vf__2657_*_*".
All wingmen immediately failed too.

Peter


ID: 3506 · Report as offensive    Reply Quote
robert

Send message
Joined: 6 Dec 07
Posts: 3
Credit: 0
RAC: 0
Message 3507 - Posted: 6 Dec 2007, 22:45:47 UTC

The VF runs are testing variable fragment sizes, ranging from 3 to 25 mers instead of the 3 and 9 mers traditional rosetta abinitio uses.

The "Incorrect fragment size requested for omega alignment. Expected 1,3, or 9, but actually got: 16" error is what the old version of rosetta used to say whenever it encountered a fragment sized outside of the norm. This was fixed for 5.85, but unfortunately when the ralph versions were updated these changes were not properly applied to the linux specific executable.

That has now been fixed, so we certainly don't expect to see that error again.

On the other hand the more recent BEST25_VFSCORE3 errors were entirely my fault. Evidently even when rosetta doesn't need 3mers for abinitio it still checks to see if they exist, and fails if they don't. Some of these runs don't use 3mers at all, so I thought I could save people some space by leaving them out of the jobs. Now, this is a mistake we would normally catch on our local machines. Unfortunately I ended up doing my tests with 3mers present anyway and didn't catch the problem before sending it to ralph. I managed to remove the jobs once the first error messages started coming back, but by that point at least a thousand were already in progress.

I definitely apologize for wasting your computational time on this, and to anyone affected, thank you for helping to catch my mistakes before they went to boinc.
ID: 3507 · Report as offensive    Reply Quote
BigMike
Avatar

Send message
Joined: 23 Feb 06
Posts: 63
Credit: 58,730
RAC: 0
Message 3508 - Posted: 6 Dec 2007, 23:35:41 UTC - in response to Message 3507.  

I definitely apologize for wasting your computational time on this...

Not a problem as far as I am concerned.

And thank you for explaining what's happening. It makes a difference to know someone is actually reading these posts. And I enjoy hearing about the nuts and bolts of Rosetta :)

==Mike

Don't believe everything you think.
ID: 3508 · Report as offensive    Reply Quote
BigMike
Avatar

Send message
Joined: 23 Feb 06
Posts: 63
Credit: 58,730
RAC: 0
Message 3509 - Posted: 6 Dec 2007, 23:40:56 UTC
Last modified: 6 Dec 2007, 23:42:45 UTC

It seems version 5.85 now has its very own message thread

See you over there...

==Mike
Don't believe everything you think.
ID: 3509 · Report as offensive    Reply Quote
Profile Conan
Avatar

Send message
Joined: 16 Feb 06
Posts: 364
Credit: 1,368,421
RAC: 0
Message 3511 - Posted: 7 Dec 2007, 21:36:30 UTC - in response to Message 3507.  

The VF runs are testing variable fragment sizes, ranging from 3 to 25 mers instead of the 3 and 9 mers traditional rosetta abinitio uses.

The "Incorrect fragment size requested for omega alignment. Expected 1,3, or 9, but actually got: 16" error is what the old version of rosetta used to say whenever it encountered a fragment sized outside of the norm. This was fixed for 5.85, but unfortunately when the ralph versions were updated these changes were not properly applied to the linux specific executable.

That has now been fixed, so we certainly don't expect to see that error again.

On the other hand the more recent BEST25_VFSCORE3 errors were entirely my fault. Evidently even when rosetta doesn't need 3mers for abinitio it still checks to see if they exist, and fails if they don't. Some of these runs don't use 3mers at all, so I thought I could save people some space by leaving them out of the jobs. Now, this is a mistake we would normally catch on our local machines. Unfortunately I ended up doing my tests with 3mers present anyway and didn't catch the problem before sending it to ralph. I managed to remove the jobs once the first error messages started coming back, but by that point at least a thousand were already in progress.

I definitely apologize for wasting your computational time on this, and to anyone affected, thank you for helping to catch my mistakes before they went to boinc.


Thanks for the update. Not a real problem I suppose as this is an Alpha project but I was wondering why it took 3 days to respond. I started reporting these problems back on the 3rd.
No worries, the only harm done is that I wont be able to do any more Ralph jobs for awhile due to all the dozens of errors (I had well over 50 of them), Boinc will not request any more work, and when it does it will only ask for a few jobs till I get successful ones returned.
ID: 3511 · Report as offensive    Reply Quote
Previous · 1 · 2

Message boards : RALPH@home bug list : Bug reports for 5.84



©2024 University of Washington
http://www.bakerlab.org