Le 14/07/2010 17:48, 顾小波 a écrit :
> [...]
> we use gpAdminIII 1.10.3 with Greeenplum database 3.3.5.0 on Solaris X64
> 10u7, one of our databases has 36 schemas, one of which has 716 tables,
> and the whole database size is about 4~5T.
Would it be possible to get a copy of your schema? (not the data of
course, but the objects definition) The idea is to restore them on my
computer, and fix the performance issue you have.
> When users open pgAdminIII
> and click the database node under the Databases node it takes a long
> time for pgAdminIII to “restoring previous environment….”, when users
> click one of the schemas, it also takes a long time for pgAdminIII to
> “retrieving Schema details…”, and the whole Interface is haling, I think
> this is due to this operation is single threaded. Another suggestion is
> can you improve gpAdmin to retrieve as little detail information as
> possible, just to retrieve it when user really need it, especially for
> high cost operations such as select count(*) from table.
>
Yes, it's single threaded. There's not a lot of count(*) uses, mainly on
system catalogs. There's one on the data (which would cause you bad
issues on a 4TB database), but it's only launched if you have bad
statistics.
Regards.
--
Guillaumehttp://www.postgresql.frhttp://dalibo.com