I have a process running on my AIX 5.3 server box. The process runs fine for 5-6days but then crashes. The log file shows malloc failure and the svmon (Virtual memory size), ps -lef (SZ value) are also gradually increasing. But unfortunately MALLOCDEBUG and any other memory debugging tools does not show any leak in source code. I am not getting the reason of this process crash. I request all the experts here in AIX to help me solve this issue.
That process you have identify .... is it from the AIX OS ?? or is Oracle ?? DB2 ?? What is it ?
Is nice to find out such errors by analizing core files and even dump ones from your server but it might be dangerous if your appl is on production. Place a call to IBM and send them all necessary files, they have tools to analize and a big database where to look for such problems. I usually ask for the current version of the AIX system:
oslevel -s
and apply the latest available maintenance level, as well as apply latest maintenance level of the failing appl.
is the process you're talking about a java process - and you rather have shared memory issues than real ones? There's a known issue with shared memory hungry 32bit applications like websphere, mq and contentmanager on 64bit AIX - and IBM has a few patches available that can help you depending on the application ...
Thank you all for replying. The process that is crashing is of my own application, written in C. The application is on production and what all I found from core files and dumps has already been analyzed. But unfortunately it din't help. The latest maintanance patch is already being used on server AIX5.3. Please let me know what I can do to avoid this crash. I have analyzed the Heap area continously growing for this process. But all allocations and deallocations has been done at required points. Please suggest me.