valgrind and some minutes watching memoryusage of owserver is still showing a contantly growing memory usage.
output is here: http://dgx.de/temp/owdebug3.2009.03.31.tar.bz2 ...the memory-leak would be no problem in a "normal" use case, but for measuring/controlling often low-end-computers are used, and mosty they are running months or years without reboot, so problems can arise from it. i guess a periodic restart of owserver can affect reliability (detection of the usb-adapter sometimes needed some tries). the long-test of the previous cvs-version used 297MB rss after 24 hours. i will long-test the current cvs-version too. > -----Ursprüngliche Nachricht----- > Von: "Paul Alfille" <paul.alfi...@gmail.com> > Gesendet: 31.03.09 20:03:58 > An: owfs-developers <owfs-developers@lists.sourceforge.net> > Betreff: Re: [Owfs-developers] owfs memory consumption and timing questions Found th bug (actually my last fix added another leak.) > > 2009/3/30 David Schumacher <davids...@web.de> > hi, > > yes, owserver uses less systime than the older version (it already > used very little cpu before). > > unfortunately it seems there is still a leak. > > here is another valgind and memory-debug-output: > http://dgx.de/temp/owdebug2.2009.03.30.tar.bz2 > > i started another long running test again. > > unfortunately i cant easily open up access to a machine with a DS2413 > attached at the moment. > if its usefull / if you would like to test by yourself, i will make a > fresh installed machine with a connected DS2413 accessible. > > david > > > -----Ursprüngliche Nachricht----- > > Von: "Paul Alfille" <paul.alfi...@gmail.com> > > Gesendet: 29.03.09 03:50:41 > > > An: owfs-developers@lists.sourceforge.net > > Betreff: Re: [Owfs-developers] owfs memory consumption and timing > questions > > Very helpful. I found the memory leak. Can you test the current CVS > > version? (I also added a bit of efficiency to this path, which needs > > testing). > > > > Paul > > > > On Sat, Mar 28, 2009 at 6:31 PM, David Schumacher <davids...@web.de> > > wrote: > > hi, > > > > i uncommented "#define OW_ALLOC_DEBUG" in module/owlib/src/include/ > ow_ > > alloc.h (i think you meant that) and ran owserver with the owwrite- > > loop for a while. > > the resulting mem.txt is still 200Kb when bzipped, too big for this > > list. > > i have no clue what to read from the file, so it can be downloaded > > from http://dgx.de/temp/owdebug.2009.03.28.tar.bz2 . > > > > the results from a valgrind-check (owserver-valgrind.txt) and an > > owwrite-loop running for 22 hours (simple-write-test.txt) are in the > > archive too. > > > > greets, > > david > > > > p.s.: > > i used the usb-adapter (DS9490R) again to be able to do more writes > > per time. > > with that adapter, the device-directories are still missing > sometimes > > when doing owdir, but writes are still arriving the 1wire-devices ( > > the LEDs on DS2413 are still blinking) during the absence of the > > device-directories... > > > > > -----Ursprüngliche Nachricht----- > > > Von: "Paul Alfille" <paul.alfi...@gmail.com> > > > Gesendet: 28.03.09 20:11:07 > > > > > An: owfs-developers@lists.sourceforge.net > > > Betreff: Re: [Owfs-developers] owfs memory consumption and timing > > questions > > > > On Fri, Mar 27, 2009 at 3:20 PM, David Schumacher <davids...@web.de> > > > wrote: > > > hi, > > > > > > i compiled the latest cvs-version [updated 2009-03-27 16:13] by > > > removing libtool from owfs-directory and symlinking to /usr/bin/ > > > libtool [which is 2.2.4]. > > > > > > sorry, owserver is still growing when writing to DS2413. see > > > attachment. > > > the first pmap ist before the start of the following loop: > > > > > > Ok, the current CVS version has some debugging enabled for memory > > > allocation/free. > > > > > > In file module/owlib/src/h/ow_alloc/h line 34 uncomment to #define > > OW_ > > > ALLOC_DEBUG > > > > > > run the program in the foreground piped a text file. e.g. > > > > > > /opt/owfs/bin/owhttpd --foreground -u -p 4444 > ~/mem.txt > > > > > > This will create a text file of all allocations and frees by > > location. > > > I sort it to match up and see if there is any memory leak. So far > I > > > haven't found any problems, but I don't have as much writing, or > > > DS2413s to truly test your findings. (You can just send me the > file. > > > Basically I use standard unix utilities, like sort and uniq -c to > > > make sense of it.) > > > > > > Paul Alfille > > > > > > ------------------------------------------------------------------ > -- > > -- > > > > > --------_______________________________________________ Owfs- > > > developers mailing list Owfs-developers@lists.sourceforge.net > https: > > // > > > > > > lists.sourceforge.net/lists/listinfo/owfs-developers > > > > -------------------------------------------------------------------- > -- > > -------- > > _______________________________________________ > > Owfs-developers mailing list > > Owfs-developers@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/owfs-developers > > > > -------------------------------------------------------------------- > -- > > --------_______________________________________________ Owfs- > > developers mailing list Owfs-developers@lists.sourceforge.net https: > // > > > > lists.sourceforge.net/lists/listinfo/owfs-developers > > ---------------------------------------------------------------------- > -------- > _______________________________________________ > Owfs-developers mailing list > Owfs-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/owfs-developers > > ---------------------------------------------------------------------- > --------_______________________________________________ Owfs- > developers mailing list Owfs-developers@lists.sourceforge.net https:// > > lists.sourceforge.net/lists/listinfo/owfs-developers ------------------------------------------------------------------------------ _______________________________________________ Owfs-developers mailing list Owfs-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/owfs-developers
