cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Enke, Michael" <michael.e...@wincor-nixdorf.com>
Subject Re: 2: Please help! Caching or Threadproblem or what else?
Date Mon, 02 Sep 2002 10:15:45 GMT
"Enke, Michael" wrote:
> 
> Vadim Gritsenko wrote:
> >
> > Enke, Michael wrote:
> >
> > >I tried it now under Windows ME:
> > >The swap file C:\WINDOWS\WIN386.SWP
> > >was growing until C:\ was full (swap file size about 500 MB) :-(
> > >
> >
> > But this means that this is not related to Java - Java's total memory is
> > limited to ... whatever you specify.
> >
> > Or this can be bug in JDK (less possible). What do you think?


here I post the problematic file. I let it run from command line after setting CLASSPATH:
java -Djdbc.drivers=$my_DB_driver_class org.apache.cocoon.Main -c . -u ERROR test801.xml
(no further transformation).

Interesting are only the last few lines where I produce the mass data
(for loop with createRow method). I print the rownumber to System.out.

1) I changed the namespace for esql logicsheet so that esql is not applied:
   At line 1271 there is a one second delay,
   after this it is going on. That's probably the gc? I don't know.
   The memory consumption is ok (20 ... 25 percent).

2) With correct namespace, when esql is applied, it blocks at rowNumber 1271,
   sometimes for one minute, sometimes forever.
   I use only one sql statement: String SQL_QUERY="select 0,0" (I have postgres)
   so you can easily change for other DB system, e.g. select 0 from dual (for Oracle)
   I watch the memory usage with the command "top" (on linux), delay between updates
   1 second, sort by memory usage. The percent MEM is: 20% ... 30% ... 40% ...
    50% ... 82% ... 42% ... 60% ... 70% ... 80% ... 80% ... 80% ... always 80%
   I have 320MB physical memory and 264MB swap space.
   StoreJanitor settings are the default values from cocoon.xconf

If I cut some code before the for loop (which has no relevance to the loop),
the blocking disappears, than there is only delay of 1 second ... 1 minute,
depends how mutch code is removed.

Maybe this a problem of StoreJanitor?

I appreciate for any help.

Michael


Mime
View raw message