Re: Document Server Memory Leak Suspicion

Fernand Vanrie <[email protected]>
Newsgroups gmane.comp.openoffice.devel.api
Message-ID <[email protected]>
  Chris ,
>  Thanks for the comment/insight... I'll try to investigate, but not 
> hopeful.
>
> It happens once a document is closed, seemingly (could be on open, 
> too)... and each subsequent/repeated opening of the file takes longer 
> and longer. It really just seems like the memory isn't being cleared 
> on close, is my best guess.
The memory is not handled by the document, but is suppose every object 
can cause a leak, in fact only closing OO itself is clearing all the 
used memory. So trie to find witch objects are leaking. I found a lot of 
emprovements in the latest versions, did you trie to use the latest 3.2 
or 3.3 ?

  For the record, the ImageControls are no longer leaking sinds 3.2 :-)

Fernand




>
> Chris
>
>
> On 11/02/2010 11:12 AM, Fernand Vanrie wrote:
>> Chris ,
>>
>> I have no experience with server applications but in the past, i 
>> found some "leaks" when using "ImageControls", its not necerarly the 
>> document itselfs who leaks, Trye to reproduce the operation on a 
>> local machine ad see when or where the memory eating starts ?
>>
>> Fernand
>>>  I hope this is an active mailing list ~ we have a serious issue 
>>> with the document server and are trying to track down some more 
>>> knowledge that might exist about it.
>>>
>>> I have new information from further testing, but here's a quick 
>>> summary:
>>> We have a program that calls the document server to load a PPT file 
>>> (as explained below). That goes fine, and then we call it to close 
>>> the file after some time... also goes fine -- mostly. The file will 
>>> close, but then the memory continues to be eaten up by document 
>>> server until the machine becomes non-responsive.
>>>
>>> We've run strace and can't find anything.
>>> We're calling the server over port 2803.
>>> We're calling the close method (which I've learned also calls 
>>> dispose, etc.).
>>>
>>> We've got hundreds of products at customers and some are beginning 
>>> to doubt OpenOffice as a platform... please help!
>>> Chris
>>>
>>> -------- Original Message --------
>>> Subject:     Document Server chaos -- not talking properly, 
>>> intermittently
>>> Date:     Tue, 26 Oct 2010 15:46:23 -0400
>>> From:     Chris Rider <[email protected]>
>>> Organization:     MessageNet Systems, Inc.
>>> To:     _OpenOffice API Developers List <[email protected]>
>>>
>>>
>>>
>>>  (I previously posted something about Document Loader - that is now
>>> revised and different, hence this post ~~ You'll find this post much
>>> more informative, I hope!)
>>>
>>> Please forgive me for causing any confusion... I was put on this 
>>> project
>>> after another engineer here took another position, so I've been playing
>>> catch-up on understanding it fully enough to at least put out a lucid
>>> question to this community. Your patience and time is very much
>>> appreciated!!!
>>>
>>> The previous developer wrote a C++ program that we call DocumentLoader,
>>> that talks to the Document Server, and loads a PPT file. The Loader 
>>> file
>>> has been through extensive testing and debugging, so I/we HIGHLY doubt
>>> the problem lies within it (been wrong before though!).
>>>
>>> Typically, these PPT files are setup as "slideshows" that auto-advance
>>> their slides and usually repeat/rotate, as managed by our own 
>>> product (a
>>> type of multimedia messaging solution). Just for the sake of example,
>>> this might be a "daily announcements" slide-show set to display 
>>> (amongst
>>> other types of messages) every 5 minutes (the other types of messages
>>> would show the remainder of the time, rinse, repeat, etc.)
>>>
>>> So here's the problem......
>>> Whenever DocumentLoader tells the Document Server to take down the PPT
>>> file, it appears to; but when it comes time to load it to display it
>>> again, it will - seemingly without problem... except that the machine's
>>> memory keeps getting sucked down (the machine that's running the server
>>> program). And, each time it cycles through (loads), there is about
>>> one-second added to the load time. So a PPT that normally takes about 2
>>> seconds to load, by the 6th time, takes like 8 seconds. And to make
>>> matters worse, once we hit about a dozen or so loads, the machine
>>> practically becomes non-responsive until we restart it.
>>>
>>> Also, (more just to see my debugging logic) we occassionally see
>>> multiple instances of the DocumentLoader program (like 20 or more)
>>> trying to load the same PPT file. I figure this is "normal"
>>> in-so-much-as the loader is trying to talk the server into loading that
>>> PPT (in the timed loop I described above), and the server is failing 
>>> for
>>> whatever reason to either respond to the loader and/or actually load 
>>> the
>>> file.
>>>
>>> I've been consulting with the previous engineer, as well as another
>>> programmer here, and there is some agreement that there could be a
>>> memory leak in the server program. It is very hard to debug, because we
>>> have this deployed on hundreds of machines, and only a handful or so
>>> behave this way ~ seemingly randomly, at that.
>>>
>>> I've not been able to find much info about this "server" ~~ can anyone
>>> enlighten me?
>>> Furthermore, is this a known issue of any kind?
>>>
>>> Even if just to say "welcome" (I'm new to the list), I'd appreciate any
>>> ideas!!!  :)
>>>
>>> Humbly,
>>> Chris
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: [email protected]
>>> For additional commands, e-mail: [email protected]
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [email protected]
>> For additional commands, e-mail: [email protected]
>>
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.