Re: Is a Multithreaded Ingres On the Horizon???

In article <26ebh1INN4n2@ariel.ucs.unimelb.EDU.AU>, richard@asis.unimelb.edu.au (Richard Begg) writes: > perl@dwrsun4.UUCP (Robert Perlberg) writes: > > |There could be other advantages besides prioritizing. I have found > |that a single Ingres server can only seem to handle 2 queries at a > |time. We have had situations where 2 big queries (big being defined as > |queries which take many minutes to run) were running on the same server > |process at the same time. During this time, no other users could start > |any queries on this server and users just trying to connect could not > |connect. ...... > > You probably could have solved this problem by playing with the > -quantum parameter in rundbms.opt. > The operative word above is "probably". We've been living with this problem from the day we first started using INGRES 6. Manipulating the quantum parameter is one of many many solutions suggested by INGRES; none of them have worked. What HAS ameliorated the situation somewhat has been successive versions of 6: the number of episodes and the range of applications that cause the episodes have decreased. This suggests that there are still paths in the code that are not behaving properly in a "multithreaded" environment. Something that would be a big help would be the capability to set job priorities. greg pavlov pavlov@fstrf.org