complain about closing bugreports with wrong reason - Mailing list pgsql-admin

From splarv@ya.ru
Subject complain about closing bugreports with wrong reason
Date
Msg-id 7af72094-8bd6-4fe0-8578-0badcb237c6d@ya.ru
Whole thread Raw
Responses Re: complain about closing bugreports with wrong reason
List pgsql-admin
Hi all.
There is a bug in PgAdmin, if I use an option "set role" in the 
connection dialog, I will not be able to cancel query anymore.

How to recreate. Config connection to a server to use "set role": the 
dialog "properties" of a server, tab "connection", field "Role" assign 
to DBO, without superuser privilege.
Run query, for instance "select pg_sleep(10000)" and try to finish it.

You will get
ERROR:  permission denied to cancel query
DETAIL:  Only roles with privileges of the role whose query is being 
canceled or with privileges of the "pg_signal_backend" role may cancel 
this query.
STATEMENT:  SELECT pg_cancel_backend(497843);

First bug in the issue tracker is:
https://github.com/pgadmin-org/pgadmin4/issues/7442

The answer to assign user to the role pg_signal_backend is wrong and 
will rise even security issue, the user will be able to terminate not 
only his query, but everyone.

I reopened this bugreport for this reason.
https://github.com/pgadmin-org/pgadmin4/issues/7821

The answer was just lie:

The connection made using the details from server dialog is used 
everywhere. pgAdmin cannot create a connection separately without the 
role specified and execute pg_cancel_backend.
This is not a bug.
This behaviour is same as psql.
End cite.

The lie because a psql and DBeaver worked perfectly in this situation. 
Everyone can set role in DBeaver connection dialog or in the psql "set 
role dbo" after connection, run query and cancel query. This doesn't 
work only in the PgAdmin and only due to reason, because developers 
simply close bugs instead on fixing it.

If your way do nothing with buggy behavior of "set role", then just 
remove it from the connection dialog.
-- 
Olleg




pgsql-admin by date:

Previous
From: Peter Geoghegan
Date:
Subject: Re: Is index deduplication active on an index
Next
From: Sathish Reddy
Date:
Subject: Pg partman