On Wednesday 04 October 2006 06:36, a.mitani@xxxxxxxxxxxxxx wrote:
> Hi,
>
> Last night, I found a problem in dynamic recovery module.
>
> I have run "pgbench" during recovery restart with 3 cluster DBs.
> In some case, the dynamic recovery was freezed.
>
> At last, I found the cause of this issue.
> I'm going to test the this recovery module patch tonight...
>
> Sorry for late release.
We should be thanking you for this product.
so...Thanks!
> -------------------------
> At.Mitani
>
> > I just wanna ask you as promised on Monday!
> > Did you patch the version 1.3 (Postgres 8.0) of PGCluster ?
> > If you did not schedule it for today : is it possible to give me the
> > patch and show me how to use it?
> >
> > Thanks a lot !
> > Largo
> >
> > PS : My boss is putting me under pressure : he is asking me everyday for
> > a solution!
> >
> > -------------------------------------------------------------------------
> >--------------------
> >
> > Largo schrieb:
> >> Thanks for the answer
> >>
> >> I though U fixed it for the other versions.
> >> I am working on 7.4.7 or really now for my new applications : 8.0 .
> >> And for these versions DID YOU PATCH IT ALSO ;
> >> I will ask you again on Wednesday .
> >> Have a nice day .
> >>
> >> Largo
> >>
> >> ------------------------------------------------------------------------
> >>
> >> a.mitani@xxxxxxxxxxxxxx schrieb:
> >>> Hi,
> >>>
> >>> I beleive that the dead-lock issue was solved in the latest version
> >>> which
> >>> is released last Friday.
> >>> The PGCluster-1.0.12rc1 is based on PostgreSQL7.3.5.
> >>> I'm going to make same patch for the each version of PGCluster.
> >>>
> >>> It takes a few days to create patch and test for the each version.
> >>>
> >>> Regards,
> >>> ----------------
> >>> At.Mitani
> >>>
> >>>> I just wanted to remind you of the PATCH concerning the problems of
> >>>> PGCluster.
> >>>>
> >>>> Is the patched version already released.
> >>>> Where can we download it if it is possible ?
> >>>>
> >>>> Thanks .
> >>>> Largo
> >>>>
> >>>> Largo schrieb:
> >>>>> Hi
> >>>>>
> >>>>> I will wait for your new version of pgcluster , please notify me if U
> >>>>> distribute it.
> >>>>> Iエll wait until Monday ; otherwise I must present another solution
> >>>>> for my Boss based on an asynchronous replication, what I don t like !
> >>>>>
> >>>>> I hope we can get this problem resolved ...
> >>>>>
> >>>>> Waiting to read from U ,
> >>>>> Largo .
> >>>>>
> >>>>> ----------------------
> >>>>>
> >>>>> a.mitani@xxxxxxxxxxxxxx schrieb:
> >>>>>> Hi Largo,
> >>>>>>
> >>>>>> I'm sorry about late response.
> >>>>>> I already had received some bug reports of the lock confliction
> >>>>>> problem.
> >>>>>> And I also received the patch of it.
> >>>>>>
> >>>>>> Therefore, in these weeks, I installed those patches and tested.
> >>>>>> I'm going to release next version which is fixed those bugs in this
> >>>>>> weekend.
> >>>>>> In the next version, it is implemented some new function which is
> >>>>>> concerning with the recovery as follow.
> >>>>>>
> >>>>>> - The queue which put the received query during recovery changed to
> >>>>>> file
> >>>>>> from memory (message queue).
> >>>>>> - Not only "rsync" but "pg_dump" can be chosen as the recovery
> >>>>>> method.
> >>>>>> (This means that PGCluster is dynamically upgradable. )
> >>>>>>
> >>>>>> Regards,
> >>>>>> ------------
> >>>>>> At.Mitani
> >>>>>>
> >>>>>>> Did someone tested intensively the PGCluster project.
> >>>>>>>
> >>>>>>> I found through my tests an unaccepted behaviour concerning some
> >>>>>>> important aspect of a database.
> >>>>>>>
> >>>>>>> - With concurrent transaction updates :
> >>>>>>> The RP does not give a response an repearts the queries and logs a
> >>>>>>> *"LOCK conflict"*, no one will be able to work on the cluster-DBs .
> >>>>>>> - *Deadlock* will not be recognized so that the system cancels a
> >>>>>>> transaction: the RP must be restarted and the SQL operations must
> >>>>>>> be
> >>>>>>> killed.
> >>>>>>> - Random() values will be differently generated on the different
> >>>>>>> components of the cluster (the 2 cluster-DBs) because they won t be
> >>>>>>> executed on a cluster first and then copied; but will be executed
> >>>>>>> parallelely on the cluster-DBs and so we will work with different
> >>>>>>> values
> >>>>>>> generated by *random()*.
> >>>>>>>
> >>>>>>> - What about *triggers* or *functions* ? Are they equally
> >>>>>>> replicated
> >>>>>>> on
> >>>>>>> both cluster-DBs or no!
> >>>>>>>
> >>>>>>> Did someone observe some of these events?
> >>>>>>> Is there a solution to get the performance of the PGCluster system
> >>>>>>> higher
> >>>>>>> !
> >>>>>>> Because I think if these points are not solved ; PGcluster will not
> >>>>>>> be
> >>>>>>> mature to deploy as a productive system using by serious
> >>>>>>> DB-applications
> >>>>>>> ...
> >>>>>>>
> >>>>>>> Waiting for REPLY !
> >>>>>>> Largo.
> >>>>>>>
> >>>>>>> PS : I am waiting since 2 days of an answer !
> >>>>>>> _______________________________________________
> >>>>>>> Pgcluster-general mailing list
> >>>>>>> Pgcluster-general@xxxxxxxxxxxxx
> >>>>>>> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> Pgcluster-general mailing list
> >>>>>> Pgcluster-general@xxxxxxxxxxxxx
> >>>>>> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >>>>>
> >>>>> _______________________________________________
> >>>>> Pgcluster-general mailing list
> >>>>> Pgcluster-general@xxxxxxxxxxxxx
> >>>>> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >>>>
> >>>> _______________________________________________
> >>>> Pgcluster-general mailing list
> >>>> Pgcluster-general@xxxxxxxxxxxxx
> >>>> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >>>
> >>> _______________________________________________
> >>> Pgcluster-general mailing list
> >>> Pgcluster-general@xxxxxxxxxxxxx
> >>> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >>
> >> _______________________________________________
> >> Pgcluster-general mailing list
> >> Pgcluster-general@xxxxxxxxxxxxx
> >> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >
> > _______________________________________________
> > Pgcluster-general mailing list
> > Pgcluster-general@xxxxxxxxxxxxx
> > http://pgfoundry.org/mailman/listinfo/pgcluster-general
>
> _______________________________________________
> Pgcluster-general mailing list
> Pgcluster-general@xxxxxxxxxxxxx
> http://pgfoundry.org/mailman/listinfo/pgcluster-general
|
Try Searching:
servers, voip, java, networking, microsoft ...
|
|
|
| 
