Posts by Trog Dog |
|
1) Message boards : :
Number crunching :
All wu crash except classical...
Posted 1992 days ago by Trog Dog Any ideas about these ones Mark? http://boinc.gorlaeus.net/result.php?resultid=7227744 http://boinc.gorlaeus.net/result.php?resultid=7242598 http://boinc.gorlaeus.net/result.php?resultid=7250864 |
|
2) Message boards : :
Number crunching :
Code 3 Errors
Posted 2057 days ago by Trog Dog [rant]stupid Akismet, now I lost my post![/rant] Cheers - that thread shows that the reason for the code 3 errors is unknown, and there's not much that we crunchers can do to help debug. As to my 6.1.0 clients they are compiled from trunk - trying to compile from 5_10 results in make errors, compiling from trunk doesn't (at least in linux) and I've encountered no errors or problems from using this version. |
|
3) Message boards : :
Number crunching :
Code 3 Errors
Posted 2062 days ago by Trog Dog What's with the code 3 errors on the trajtou-cu111 wu's. The odd random host seems to crunch it successfully - all other hosts error out. |
|
4) Message boards : :
Number crunching :
More Weird WU's
Posted 2183 days ago by Trog Dog Have a look at this wu. So far there are 12 replications. 9 of which are to the same host - 15035 and 2 to host 16635. |
|
5) Message boards : :
Number crunching :
All WU's in my (client) buffer are in the state 'Client Error'
Posted 2185 days ago by Trog Dog G'day kdr I'm talking about exactly the same problems that you are. WU's are being listed as Client Error before they are even returned - this is causing the LC server to reissue another copy of that wu to a new host - and if that host returns the result before you do, when you return the result you get no credit and result is marked as too late. This is an ongoing problem, and I appreciate that this is a beta project, all I want to know is that a) the problem is acknowledged as being a problem, and b) it is either being worked on or we just have to live with it. If we just have to live with it then I'll continue aborting those affected wu's (when I notice in time :) ) |
|
6) Message boards : :
Number crunching :
All WU's in my (client) buffer are in the state 'Client Error'
Posted 2186 days ago by Trog Dog This isn't a new problem and it isn't related to the BOINC version that is used. The problem is also reported in this thread. It probably also explains the problems in this thread too. Below is an example - Result ID 5535188 Name wu_81185150_1180096778_1889_1 Workunit 2058635 Created 26 May 2007 23:29:29 UTC Sent 26 May 2007 23:29:56 UTC Received --- Server state Over Outcome Client error Client state New Exit status 0 (0x0) Computer ID 22665 Report deadline 2 Jun 2007 23:29:56 UTC CPU time 0 stderr out Validate state Initial Claimed credit 0 Granted credit 0 application version --- This wu has been marked as an error immediately when it was sent to me - it has already been sent to another host. If I were to crunch this unit and return it would be marked as result returned too late - I won't be crunching it though - whenever I come across these types of wu's I abort them. This does not happen with every wu but happens regularly enough to be be a problem. |
|
7) Message boards : :
Leiden Classical :
real AMD 64bit apps ??
Posted 2190 days ago by Trog Dog Have you run 'ldd' on the respective executables to check for missing libraries? |
|
8) Message boards : :
Leiden Classical :
real AMD 64bit apps ??
Posted 2192 days ago by Trog Dog Sweet. My 64bit boxes need another project to keep them busy :) |
|
9) Message boards : :
Number crunching :
Client state showing error even though WU still in progress
Posted 2210 days ago by Trog Dog This is still occurring. I'm currently aborting any wu's that are showing this state - which is far from ideal. |
|
10) Message boards : :
Number crunching :
WUs error out on AMD64 with Linux
Posted 2312 days ago by Trog Dog I am running two machines with 32 bit Debian linux. I am getting similar errors. If you install freeglut3 and libstdc++5 that should fix both these problems. |