On 15/12/11 14:17, Bill Roberts wrote:
On 15 Dec 2011, at 13:43, Andy Seaborne wrote:
You can't run tdbdump on a database that Joseki or Fuseki is using. One JVM at
a time only.
Hmm, that's interesting. I quite often run tdbdump on a TDB that is also being
used by a running Joseki. I haven't had any problems so far - ie it works. Is
there a risk of this corrupting the TDB database?
With tdbloader I generally stop Joseki first - but also interested to know the
locking situation there: how dangerous is it to run tdbloader and joseki/fuseki
simultaneously?
Well, it's possible with very careful use and knowing what's going on
although with a reasoner involved, I'm not sure what happens exactly.
IF the database is sync'ed and so consistent
AND you don't touch anything
THEN tdbdump will work currently.
BUT it's never guaranteed
IF your server is merely publishing (read-only)
THEN it's currently OK.
You absolutely must stop the server to use tdbloader.
Oddly, there is chance all that breaks are queries and the DB is valid
and OK (i.e. the server caches are inconsistent with the DB).
What is needed is a Fuseki control panel and a "dump" button along with
switchover and "make read-only" and other things.
Some of these are likely as they are things Epimorphics would find
useful; other things need resourcing (funding) or they will take a long
time to bubble up the priority stack on my non-Epimorphics time.
Patches help.
Andy
Thanks
Bill
