Subject | Re: [firebird-support] Firebird 1.0.2 Fails when tickcount rolls over |
---|---|
Author | Milan Babuskov |
Post date | 2003-09-16T14:18:25Z |
bmorrison_navco wrote:
to do with sequencing of IP packets, and at critical moment one packet
gets lower number than previous, so system concludes that connection is
dropped. Firebird is not the only application that can suffer from it.
AFAIK, the solution is to either use other OS, or to restart the
computer periodically (once per month should do it).
HTH
--
Milan Babuskov
http://fbexport.sourceforge.net
> I've been using Firebird 1.0.2 on NT 4 with one local connection andThis is (not so) well known issue with Win9x/NT. AFAIR, it has something
> three or me tcp/ip connections (typically loopback). One of these
> tcp/ip connections is typically inserting records into the database
> at the rate of 6/second.
>
> Under Windows there is an internal DWORD that contains the number of
> milliseconds since the system was turned on. This value resets
> (maxes out) after about 49.7 days and resets to 0.
>
> When this this rollover occurs, the next record insert fails with
> the following error :
>
> ISC ERROR CODE:335544648
>
> Which translates into : Connection lost to pipe server
>
> At this point the server has become unstable and responds with that
> same error to any operation including inserts, updates, creating new
> databases, etc.
>
>
> Does anyone have any ideas or thoughts on the subject?
to do with sequencing of IP packets, and at critical moment one packet
gets lower number than previous, so system concludes that connection is
dropped. Firebird is not the only application that can suffer from it.
AFAIK, the solution is to either use other OS, or to restart the
computer periodically (once per month should do it).
HTH
--
Milan Babuskov
http://fbexport.sourceforge.net