more on crashes

Paul Holcomb bf1942-linuxserver at blue.cpoint.net
Mon Feb 17 23:32:15 EST 2003


 We've been getting core files that trace to the same location every
 time the server drops.  
 
 For reference:
 #0  0x0950670a in dice::ref2::io::NetworkManager::getNetUpdate ()

 So, this time, I'm on playing and watching this happen.  And some
 point in time there is the "weird lag" that has been discussed on the
 list and then the number of players actually decreased. People jumped
 off because it was laggy probably, but at the same time, not everyone
 saw the lag either.  

 Anyway, the memory has been steadily increasing regardless of number
 of players.  Currently it has been about 30 minutes since I first
 noticed the problem and the memory has gone from about 215MB up to
 310MB of real memory used (real and resident are the same BTW)

  963 bf1942    15   0  310M 310M 16068 R    50.1 30.7  71:22  bf1942_lnxded

 Additionally CPU percentage used has gone steadily up.

 The process takes up 1MB more every minute.

 I think this section of code is a safe place to look for a memory
 leak :)

 What is this code, connection tracking? 

-- 
Paul Holcomb                               *pholcomb    \@      cpoint  net*
Sr. Network Engineer                       Counterpoint Networking, Inc.



More information about the Bf1942 mailing list