Subject | Re: [ib-support] Re: Firebird 1.0 goes boom |
---|---|
Author | Paul Fleischer |
Post date | 2002-01-27T23:18:58Z |
ons, 2002-01-23 kl. 18:49 skrev rogervellacott:
are
completelt different.
(it
ran for 30 minutes before I killed the process).
However, by killing the specific process (by ID), rather than killing of
all
interbase connections helps...
... my question is, does a garbage sweep really do this?
--
Paul Fleischer
> I don't think sweeping will stop users logging in. A more likelyThey are - and since this is on a Unix machine upper and lowercase files
> reason is...
> Users are logged in to the server on TCP/IP. (SERVER:C:\dATA ETC)
> Someone then logs in, using the server as a client, as a Local
> Server, not using TCP/IP.(c:\DATA ETC).
> The server then starts screwing up the database because it is
> treating the same file as two separate databases. This often
> prevents further logins. If you are very lucky, the lock-outs happen
> before any significant data or metadata updates are made, or you kill
> the server before any significant transactions get committed, and you
> don't corrupt the entire database.
>
> So ensure that all log-ins are using the same protocol, and precisely
> the same path to the database file.
are
completelt different.
>Well, it seems to run forever, and no interbase connections are allowed
> --- In ib-support@y..., "Richard Pendered" <rpendered@t...> wrote:
>
>
> > not wierd behavour, it is doing a garbage sweep. set the sweep
> interval to 0
> > and run sweep once a day/week with cron.....
(it
ran for 30 minutes before I killed the process).
However, by killing the specific process (by ID), rather than killing of
all
interbase connections helps...
... my question is, does a garbage sweep really do this?
--
Paul Fleischer