Is anyone having problems with clients randomling stalling??
As in the lock file exists. The Client is in memory.
But update.txt and filelist.txt are not getting updated for 10,12 or 24 hours.
Even though processor usage is around 98,99%.
I know it doesn't make sense, the first 3 or 4 times I figured I was imagining things. But now it has happened too many times
Error log is clean and the client will shutdown and restart without errors.
Anyone else??
The only thing I've had like that is where the client will terminate itself, but the lock file is still present, but it's not loaded into memory or running as yours is. Mine as well has no errors in the log at all and a simple restart fixes it (w/o having to delete anything or such).
Is your client running with Net Disabled by any chance?
I've the same problem on 3 crunchers. I call it "Silent death"Originally posted by ^7_of_9
The only thing I've had like that is where the client will terminate itself, but the lock file is still present, but it's not loaded into memory or running as yours is. Mine as well has no errors in the log at all and a simple restart fixes it (w/o having to delete anything or such).
Is your client running with Net Disabled by any chance?
take it out of quiet/service mode and run it normally to watch the console display and see if it is working
Use the right tool for the right job!
What is shown on the console should be shown in the error.log......
I've seen things on the console that don't make it into the error.log. I don't think I have seen it write to the error.log if it doesn't delete the lock file.
I don't really know much about how this stuff works, but I think that if the client crashes hard enough it doesn't write to the log or delete the lock file, but whatever process that it was calling returns an error that gets printed to the console even though the client is dead.
my point was that if you run the console, you will see the ascii picture being drawn and it will say what part it is doing and stuff
if it is really broken , the console output should make it obvious
Use the right tool for the right job!
Ok, I'll give it a try....
I run Knoppix with -qt switch on so there is not console output. I think most people run -qt as it's faster.
And when mine shutdown I hadn't touched the box for 36 hours and there was no error.log entry, no .lock file.
Agent Smith was right!: "I hate this place. This zoo. This prison. This reality, whatever you want to call it, I can't stand it any longer. It's the smell! If there is such a thing. I feel saturated by it. I can taste your stink and every time I do, I fear that I've somehow been infected by it."
Just to confirm, if the crash on the system is extremely severe, the client may not write to the error log. If you believe the client is stalling for many hours without doing any productive work at all, restart it in non-quiet mode to see what is going wrong.
Also keep in mind though that sometimes the client may simply be stuck trying to obtain the correct structure.
If it is some new bug, please report it in the bugs forum so that we can fix it ASAP.
Elena Garderman
I just found and fixed a strange bug on MacOSX - not sure where else it occurred. It seems it was getting stuck in an endless loop trying to pick a random filename (in the /tmp dir that is). For some bizarre reason, it kep picking the same one (which already existed)! It is supposed to move on and try a different one but didn't. So this could result 99% CPU usage but no progress. This occurs just at the start of creating a new trajectory distribution at the end of a generation. I fixed it by adding a further random number to the filename to help increase the entropy. Let me know if you still get this stalling problem in the next release or if your stalling occurs at a point other than that mentioned above.
Howard Feldman
That sound like what I was seeing.
The monitor either showed it at 46 or 1. So that would be about where you are talking about.
Glad you got it fixed!
There wasn't really enough info to prove it was really happening.
Posting Permissions
Reply With Quote