Subject | Re: [IBO] Crash |
---|---|
Author | mikko.nieminen@vvo.fi |
Post date | 2001-04-11T14:24:36Z |
Hi
Server is CitRix metaframe 1.8. At same server we have
still some heavy BDE applications. At the end and beginnig
of month where n. 200 user connected the server and they are using
1..3 applications which uses database server on another server.
All server has enough memory (1 GB) and storage base. we have three
different operative databases on SQL server.
When one of the (Interbase 5.6) crash Metaframe won't reset
connection and all applications hangs and even logout/login
don't do the trick. through load balancing some users can log in
again but the old process still running on on server where they
previosly were logged on.
I've changed one big database on different server but it wont
help. in interbase log we have errors like INET 10038, 10054
and well known "Virtual address. not priviledge do ..." and
we have removed all udfs from databases. Wont help.
In one operative database we have one stored procedure which
is used for calculating rent for apartements and it used VERY
HEAVILY i'm think that's the prolem for IB crash.
But if still make applications survive from database crash It
would make my day.
Mikko Nieminen
VVO asunnot
Finland
Server is CitRix metaframe 1.8. At same server we have
still some heavy BDE applications. At the end and beginnig
of month where n. 200 user connected the server and they are using
1..3 applications which uses database server on another server.
All server has enough memory (1 GB) and storage base. we have three
different operative databases on SQL server.
When one of the (Interbase 5.6) crash Metaframe won't reset
connection and all applications hangs and even logout/login
don't do the trick. through load balancing some users can log in
again but the old process still running on on server where they
previosly were logged on.
I've changed one big database on different server but it wont
help. in interbase log we have errors like INET 10038, 10054
and well known "Virtual address. not priviledge do ..." and
we have removed all udfs from databases. Wont help.
In one operative database we have one stored procedure which
is used for calculating rent for apartements and it used VERY
HEAVILY i'm think that's the prolem for IB crash.
But if still make applications survive from database crash It
would make my day.
Mikko Nieminen
VVO asunnot
Finland
--- In IBObjects@y..., "Geoff Worboys" <geoff@t...> wrote:
> > And guardian will put it back online quickly. My question
> > is how can i restore connection properly ie. enduser should
> > not really know what had happen.
>
> I believe Jason has some work related to this planned for IBOv4.
> Until then you have little choice but to close the application, or
at
> least reset everything, and then restart.
>
> > The real prolem lies on Citrix metaframe servers database
> > server crash leaves "ghost" (several) process on server
> > which could use more 80% processor time. These processes
> > have reset allways manually after crash. Technical support
> > dont't like me those days very match
>
> I'm a bit puzzled. My main (client) application runs on terminal
> server - which is the base for metaframe as you probably already
know.
> I dont understand where your CPU useage problem is coming from.
> Memory consumption I can understand but CPU? Perhaps you dont have
> enough memory on the server, causing lots of virtual memory swapping
> with resulting CPU delays/work.
>
> You are not trying to actually run IB Server on the metaframe server
> are you? I would not recommend this on anything but the very small
> installations.
>
> You can setup MTS sessions to automatically reset after
disconnection.
>
>
> Geoff Worboys
> Telesis Computing