Thanks, Marcel.
The reason why I asked this is that I happened to observe duplicate
rows (same JOURNAL_ID value but different REVISION_ID values) while
supporting someones on their dev server.
I couldn't figure out how that could happen. Perhaps they accidentally
started two servlet containers with the same JR cluster ID at the same
In principle, it should be better to have a PK on JOURNAL_ID for data
integrity purposes, not for generic indexing needs, IMHO.
For now, I can live with removing one of the accidental duplicate row
though. ;-)



On Tue, Mar 6, 2018 at 2:40 AM, Marcel Reutegger <> wrote:
> Hi Woonsan,
> IIRC this table contains one entry per cluster node, which usually means
> there are just a couple of entries. An index is therefore not necessary.
> Regards
>  Marcel
> On 05.03.18 15:48, Woonsan Ko wrote:
>> Hi,
>> I notice that LOCAL_REVISIONS table [1] is created without a unique
>> index even if the JOURNAL_ID seems to be used as a primary key in
>> org.apache.jackrabbit.core.journal.DatabaseJournal.
>> Is there any reason behind?
>> Thanks in advance,
>> Woonsan
>> [1]

Reply via email to