OWS 3.3 beta: CHOKES THE SYSTEM

    There is a mentioning of a problem in the OWS 3.* FAQ on www.olab.com.     It concerns system going wild and chewing on the system resources     ( memory ) till there is nothing left. It also says that the     description and the conditions of the problem are really vague,     so there's no cure being developed.

    Looks like I'm now one of those affected. I've installed OWS 3.3     on Sun Sparc along with the ORACLE 7323 a few days ago as a transition     from our OWS21 installation.

    Everything went fine and I was able to start the developing cycle     on one of my apps w/o any problem: everything worked just fine.     Yet it seemed to be slower thah the same apps I have on OWS 21.

    All of a sudden today after the system was booted up and I tried     to access the URL that references stored PLSQl procedure. the     system went wild:

    WRB broker spwaned PLSQL catrtidge which in it turn started     oracle{SID} (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))     process that started consuming
    more and more memory while there was no response from the web     server. When I tried to STOP and RELOAD from the Web browser,     extra oracle{SID} ( *** (PROTOCOL=seq)) process appeared and     the system went down trashing: constant swap, no swap was     left etc. This can go on and on if one tries to access PLSQL     stored procedures.     

    But, if I use the OWS and don't access PLSQL ( when     retreiving the filesystem docs ) , everything appers     to be quite normal.     

    Only when I kill these oracle{SID} processes the system gets     the memory back.

    Some of those wrbc prosesses also became zombies ( even prior to     me killing oracle{SID} processes.

    The RDBMS appears operational otherwise: I can login (pl/sql),     compile and run apps w/o anything like this happening, it only     happend when OWS 3.3 tries to access a stored procedure. The OWS     and ORACLE are on the same machine.

    I didn't patch neither the OS ( Sparc Solaris 2.5) nor the RDBMS     itself with any patches, so could it be the problem ?

    Rashid.

    E-mail copy of reply would be appreciated. Received on Wed Mar 19 1997 - 00:00:00 CET