Closed portnov closed 8 months ago
Hi Ilya!
Thank you for this report. I'll include your fixes in a script during a week or two. point 1 and 3 should be implemented, but point 2 seems to be done outside of a script - I don't think it is a good idea to bind a schema inside the script. Point 4 looks as similar binding. To address this issue there is a database exclusion list for every server in a pg_profile installation. See the _set_server_dbexclude() function at Managing servers
Thanks for quick reply! We will try using set_server_db_exclude().
Please check the current master branch, or _manual file of the last release. The last commit should fix points 1 and 3. I think point 2 (setting search_path) should be done just before script execution during install. It seems set_server_db_exclude() function should be called also during install just after the script.
@zubkov-andrei , we can confirm that the script generated from master branch is working now. Thank you!
I think this issue can be closed then.
Hello.
As far as I understood from README, it should be enough to do
make USE_PGXS=y sqlfile
to make a script, which would install pg_profile to AWS RDS. However, when we tried that, the script did not work. We had to do the following manual changes in it:Optional: Add
in the beginning just to make sure that in case of any error the script would stop instead of skipping errors.
Optional: Add
in the beginning, to install it to schema other than public. I believe this should be done almost in all cases for pg_profile.
Comment out lines mentioning ALTER EXTENSION, like
as there will be no extension in RDS.
In
collect_obj_stats
function, beforeadd
This is required, because pg_hba rules in RDS reject any connection to rdsadmin database.
Could these changes be made in pg_profile distribution, so that it would be possible to use it on AWS RDS without manual modifications?