Mailing-List: contact users-help@pdfbox.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@pdfbox.apache.org
In-Reply-To: <70874614-e0e2-968d-cb20-e253ce2ecbf0@t-online.de>
References: <OF93939964.88DD21FB-ONC125806D.005671F2-C125806D.00569CA9@id.dvag.com> <70874614-e0e2-968d-cb20-e253ce2ecbf0@t-online.de>
To: users@pdfbox.apache.org
MIME-Version: 1.0
Subject: Antwort: Re: Problems with ScratchFileBuffer and Finalizer: OutOfMemory
From: Tjard Kopka <Tjard.Kopka@dvag.com>
Message-ID: <OF1151C3AE.406337E5-ONC125806E.0034AD53-C125806E.0035C918@id.dvag.com>
Date: Thu, 17 Nov 2016 10:47:28 +0100
Content-Type: multipart/alternative; boundary="=_alternative 0035C916C125806E_="
archived-at: Thu, 17 Nov 2016 09:48:14 -0000

--=_alternative 0035C916C125806E_=
Content-Type: text/plain; charset="utf-8"
content-transfer-encoding: quoted-printable

Hello,

all the Finalizer-Instances in the heap dump have a reference to=20
ScratchFileBuffer-Instances:

class java.lang.ref.Finalizer @ 0xc0005768 System Class=20
|- java.lang.ref.Finalizer @ 0xed44ed10=20
|  |- java.lang.ref.Finalizer @ 0xed43c9b0=20
|  |- org.apache.pdfbox.io.ScratchFileBuffer @ 0xed44ecd0

where as example
        java.lang.ref.Finalizer @ 0xed44ed10=20
has as referent;
        org.apache.pdfbox.io.ScratchFileBuffer @ 0xed44ecd0


I think the reason herefore is the overriden finalize method in=20
ScratchFielBuffer:

    @Override
    protected void finalize() throws Throwable
    {
        try
        {
            if ((pageHandler !=3D null) && LOG.isDebugEnabled())
            {
                LOG.debug("ScratchFileBuffer not closed!");
            }
            close();
        }
        finally
        {
            super.finalize();
        }
    }


I found this as a possible goog explanation for our out of memory crashes:
Objects which override this method (finalize) need to called by a=20
background thread call finalizer, and they can't be cleaned up until this=20
happens. If these tasks are short and you don't discard many of these it=20
all works well. However if you are creating lots of these objects and/or=20
their finalizers take a long time, the queue of objects to be finalized=20
builds up. It is possible for this queue to use up all the memory.

Thanks
Tjard


[excerpt from memory analysis snipped]
>=20
> Nevertheless, do you have a quick fix or workaround for us?

No... from what I see, the 70,1% are not pdfbox but java itself?!

Consider updating to the current 1.8 JDK version... if the problem=20
doesn't go away, try reproducing it with files you can share.


---------------------------------------------------------------------
Deutsche Verm=C3=B6gensberatung Aktiengesellschaft DVAG
M=C3=BCnchener Stra=C3=9Fe 1
60329 Frankfurt am Main
Vorstandsvorsitzender: Andreas Pohl
Mitglieder des Vorstandes: Dr. h.c. /HLU Udo Corts, Hans-Theo Franken, Chri=
stian Glanz,=20
Lars Knackstedt, Dr. Helge Lach, Robert Peil, Dr. Dirk Reiffenrath
Aufsichtsratsvorsitzender: Friedrich Bohl
Sitz der Gesellschaft: Frankfurt am Main
Handelsregister Frankfurt HRB 15511
USt-Ident.-Nr.: DE 114 139 839
Erlaubnis- und Aufsichtsbeh=C3=B6rde nach =C2=A7 34c GewO: Stadt Frankfurt =
am Main, Ordnungsamt, Kleyerstra=C3=9Fe 86, 60326 Frankfurt am Main=20
Erlaubnis- und Aufsichtsbeh=C3=B6rde nach =C2=A7 34f GewO: IHK Frankfurt am=
 Main, B=C3=B6rsenplatz 4, 60313 Frankfurt am Main=20
Gemeinsame Registerstelle f=C3=BCr =C2=A7 34d GewO und =C2=A7 34f GewO:=20
Deutscher Industrie- und Handelskammertag (DIHK) e.V.=20
Breite Stra=C3=9Fe 29, 10178 Berlin, Telefon 0180 600585-0=20
(20 Cent/Anruf aus dem deutschen Festnetz, h=C3=B6chstens 60 Cent/Anruf aus=
 Mobilfunknetzen)=20
www.vermittlerregister.info oder www.vermittlerregister.org=20
Registernummer nach =C2=A7 34d GewO: D-LYYB-BSPX5-17=20
Registernummer nach =C2=A7 34f GewO: D-F-125-93J4-60=20
---------------------------------------------------------------------=20
--=_alternative 0035C916C125806E_=--