[mdr-users] RE : [mdr-users] Re: Loosing last transaction results in case of crash
"Olivier RICHAUD" <[email protected]> Wed, 26 Oct 2005 20:21:49 +0200
| Newsgroups | gmane.comp.java.netbeans.modules.mdr.user |
|---|---|
| Message-ID | <005701c5da5a$22085390$0a64a8c0@lauris> |
Would you support Oracle per chance? Is there any documentation for your jdbc storage? I mean, is it hard to move from MDR standard repository to JDBC repository? Thanks. Olivier. -----Message d'origine----- De : John V. Sichi [mailto:[email protected]] Envoyé : mercredi 26 octobre 2005 08:05 À : [email protected] Objet : [mdr-users] Re: Loosing last transaction results in case of crash Note that jdbcstorage implementation actually commits every transaction for real. So if you use it together with a DBMS with transaction support, you don't have to worry about losing any transactions. JVS > Subject: > Re: [mdr-users] Loosing last transaction results in case of crash > From: > Martin Matula <[email protected]> > Date: > Mon, 24 Oct 2005 15:36:18 +0200 > To: > [email protected] > > To: > [email protected] > > Message-ID: > <[email protected]> > MIME-Version: > 1.0 > Content-type: > text/plain; charset=ISO-8859-1; format=flowed > Content-transfer-encoding: > 7BIT > > > Hi Olivier, > this is normal - as designed. For the performance reasons the > transactions are cached in memory and committed to disk only when the > cache threshold is reached. So, a few transaction performed just > before the crash of the application may be lost but you are guaranteed > that the transaction will be lost as a whole (not just partially), so > the state should at any case be consistent. Martin > > Olivier RICHAUD wrote: > >> Hi all, >> >> Has anybody already experienced loosing the results of the last >> transaction in case the JVM crashes. I mean, alltough the last >> read/write transaction has been committed, if the JVM crashes or is >> stopped abruptly, changes are lost. For example, although an object >> has correctly been created and the call to endTrand performed, if the >> JVM unexpectedly stops, I would loose that creation. >> >> Regards. >> >> Olivier. >> >> >> > > > ---------------------------------------------------------------------- > -- > > Subject: > RE : [mdr-users] Loosing last transaction results in case of crash > From: > "Olivier RICHAUD" <[email protected]> > Date: > Tue, 25 Oct 2005 21:38:57 +0200 > To: > <[email protected]> > > To: > <[email protected]> > > Message-ID: > <013601c5d99b$c2bb72f0$0a64a8c0@lauris> > MIME-Version: > 1.0 > Content-Type: > text/plain; charset="iso-8859-1" > Content-Transfer-Encoding: > quoted-printable > > > Thanks. > > Simple question: How can force the system so that it flushes all non > written transactions to the disk? > > Is this per chance possible to configure this? > > Olivier. > > -----Message d'origine----- > De : Martin Matula [mailto:[email protected]] > Envoyé : lundi 24 octobre 2005 15:36 > À : [email protected] > Objet : Re: [mdr-users] Loosing last transaction results in case of > crash > > > Hi Olivier, > this is normal - as designed. For the performance reasons the > transactions are cached in memory and committed to disk only when the > cache threshold is reached. So, a few transaction performed just before > the crash of the application may be lost but you are guaranteed that the > > transaction will be lost as a whole (not just partially), so the state > should at any case be consistent. > Martin > > Olivier RICHAUD wrote: > > >>Hi all, >> >>Has anybody already experienced loosing the results of the last >>transaction in case the JVM crashes. I mean, alltough the last >>read/write transaction has been committed, if the JVM crashes or is >>stopped abruptly, changes are lost. For example, although an object has > > >>correctly been created and the call to endTrand performed, if the JVM >>unexpectedly stops, I would loose that creation. >> >>Regards. >> >>Olivier. >> >> >> > > >