Just for others who receive the same kind of stacktraces. I was able to reproduce the problem by starting the component with the aggregators multiple times.
Obviously I started accidentally multiple instances and therefore multiple camel/java processes "shared" the same aggregator-hawtdbs. So, I need to check on startup if another instance for the same environment is already running and then this problem should no more occur. If I understand the docs correctly, LevelDB does this "automatically" with a file lock that a second instance would not be able to acquire. Thanks to all Stephan ************************************************************************ Von "aabändle" bis "zwüsche" - das umfangreichste Berndeutsch-Wörterbuch im Internet: http://www.berndeutsch.ch Ausserdem: Das Blog zur Website unter http://blog.berndeutsch.ch Facebook-Seite unter https://www.facebook.com/berndeutsch Google+ Seite unter http://www.google.com/+berndeutsch On Mon, Jul 22, 2013 at 12:44 PM, Stefan Burkard <sburk...@gmail.com> wrote: > @Claus: I read there ( > http://leveldb.googlecode.com/svn/trunk/doc/index.html) that it is safe > when only my camel/java process or even the writing process dies, but it is > not safe when the whole operating system dies (power outage). Perhaps I get > it wrong? > > @Raul: I don't expect that the Camel commiters build these tools, but the > people who build the database. I just wanted to point out that I use > frameworks to avoid building all by myself. > > > ************************************************************************ > Von "aabändle" bis "zwüsche" - das umfangreichste Berndeutsch-Wörterbuch > im Internet: http://www.berndeutsch.ch > Ausserdem: Das Blog zur Website unter http://blog.berndeutsch.ch > Facebook-Seite unter https://www.facebook.com/berndeutsch > Google+ Seite unter http://www.google.com/+berndeutsch > > > On Sun, Jul 21, 2013 at 1:27 PM, Raul Kripalani <r...@evosent.com> wrote: > >> > After all I use the great Camel framework to >> avoid building general-purpose functionality like this by myself. >> >> At Apache Camel we only offer a component to interact with HawtDB, but >> we're not responsible for the development nor tooling around that project. >> Sorry. >> >> Regards, >> Raúl. >> On 18 Jul 2013 08:48, "Stefan Burkard" <sburk...@gmail.com> wrote: >> >> > Hi Claus >> > >> > I'm using camel-hawtdb 2.9.6 and (according to the classpath) hawtdb >> 1.6. >> > The fact that hawtdb has no recovery tools but I need to build them by >> > myself is bad news to me. After all I use the great Camel framework to >> > avoid building general-purpose functionality like this by myself. >> > >> > How about LevelDB? Has it better tool support? And can I already use >> > leveldb with a Camel 2.x release? I stumbled over LevelDB while >> searching >> > for recovery options for HawtDB, but I didn't find an example how to >> use it >> > with the aggregator. Is there a unittest or similar I can look at? >> > >> > Thanks >> > Stefan >> > >> > >> > >> > >> > ************************************************************************ >> > Von "aabändle" bis "zwüsche" - das umfangreichste Berndeutsch-Wörterbuch >> > im Internet: http://www.berndeutsch.ch >> > Ausserdem: Das Blog zur Website unter http://blog.berndeutsch.ch >> > Facebook-Seite unter https://www.facebook.com/berndeutsch >> > Google+ Seite unter http://www.google.com/+berndeutsch >> > >> > >> > On Wed, Jul 17, 2013 at 10:10 AM, Claus Ibsen <claus.ib...@gmail.com> >> > wrote: >> > >> > > What version of Camel and HawtDB are you using? >> > > >> > > To try to recover you would possible need to write some java code with >> > > the HawtDB API to load the corrupted file(s) and peak inside. >> > > >> > > Down the road we recommend using camel-leveldb instead of >> > > camel-hawtdb. This uses LevelDB as the store instead which is a much >> > > more mature and widespread used store. >> > > https://code.google.com/p/leveldb/ >> > > >> > > Apache ActiveMQ 5.9 offers leveldb out of the box, and is being >> > > considered as the recommended/default store over its KahaDB store. >> > > >> > > The camel-leveldb has the same functionality as camel-hawtdb and is >> > > very similar to setup. >> > > >> > > >> > > On Tue, Jul 16, 2013 at 10:12 AM, Stefan Burkard <sburk...@gmail.com> >> > > wrote: >> > > > Hi Camel users >> > > > >> > > > I have a component with 2 persistent aggregators. One receives all >> > > messages, >> > > > one only a part of them. After a lot of test runs without problems I >> > had >> > > > yesterday a serious problem with the aggregator persistence >> (hawtdb). >> > > > >> > > > I don't know yet what causes the problems, but however, problems can >> > > occur. >> > > > My problem is that I cannot recover the data from the hawtdb-files. >> > > > >> > > > Im my logs, I got first of all about 8 stacktraces like the attached >> > > > "stacktrace1.txt". The number in the error message "The requested >> page >> > > was >> > > > not an extent: 35" is growing from stacktrace to stacktrace from 35 >> to >> > > 1163. >> > > > >> > > > Then, I got some stacktraces like the attached "stacktrace2.txt". >> > > > >> > > > Finally I got A LOT of stacktraces like the attached >> "stacktrace3.txt". >> > > > >> > > > After shutting down the component gracefully, I tried to restart it, >> > but >> > > > this throws stacktraces like the attached "stacktrace-startup.txt". >> > > > >> > > > I can only start the component again if I rename the hawtdb-files so >> > they >> > > > are ignored and new hawtdb-files are created. >> > > > >> > > > This leaves me with the question: how can I recover the corrupted >> > > > hawtdb-files? I didn't found anything about this subject and if >> this is >> > > not >> > > > possible, this would be a real show-stopper. >> > > > >> > > > Thanks for any help >> > > > Stefan >> > > > >> > > >> > > >> > > >> > > -- >> > > Claus Ibsen >> > > ----------------- >> > > Red Hat, Inc. >> > > Email: cib...@redhat.com >> > > Twitter: davsclaus >> > > Blog: http://davsclaus.com >> > > Author of Camel in Action: http://www.manning.com/ibsen >> > > >> > >> > >
