Hi All,

I have a distributed database in which we have two tables registered, one of these tables is large, it has 311million rows. The other is only 2millionrows. Both are tables with page_size=32k. They are identical in structure, column names, column nulls, column defaults.

Within the ddb we then create view myview as select * from table1 union allselect * from table2

If I now do 'select count(*) from myview', the query will take about 13minutes and then generate an error:
E_QE0546 RQF detected a timeout condition.

I have sysmod-ed and optimized the source databases, and run optimizedb over the ddb.
I have ensured the correct page cache (32k) is enabled on the installation holding the ddb.

All installations are Ingres10.0.0.

Note that:
1. The optimze of the star database generated numerous messages like:
W_OP0961 Truncation occurred when assigning a character string to host language variable.
2. I have another pair of tables (also 32k page) which are much smaller, these also have a view defined on them as per above. The select works perfectly well.
Any Ideas?

Martin Bowes

Do you have a lockmode timeout in effect? If so try increasing it or
unsetting it. I found some references in an old bug description saying
that the LDB queries need to complete within the "timeout" period.
However it wasn't clear whether a) this still applies and b) whether
this was the lockmode timeout or some Star specific timeout (if so I
can't find any current references to it).



From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of
Martin Bowes
Sent: 03 June 2011 12:09
To: Ingres and related product discussion forum
Subject: [Info-Ingres] E_QE0546 in STAR Query



Hi All,



I have a distributed database in which we have two tables registered,
one of these tables is large, it has 311million rows. The other is only
2million rows. Both are tables with page_size=32k. They are identical in
structure, column names, column nulls, column defaults.



Within the ddb we then create view myview as select * from table1 union
all select * from table2



If I now do 'select count(*) from myview', the query will take about
13minutes and then generate an error:

E_QE0546 RQF detected a timeout condition.



I have sysmod-ed and optimized the source databases, and run optimizedb
over the ddb.

I have ensured the correct page cache (32k) is enabled on the
installation holding the ddb.



All installations are Ingres10.0.0.



Note that:

1. The optimze of the star database generated numerous messages like:

W_OP0961 Truncation occurred when assigning a character string to
host language variable.

2. I have another pair of tables (also 32k page) which are much smaller,
these also have a view defined on them as per above. The select works
perfectly well.

Any Ideas?



Martin Bowes