Обсуждение: pg_ctl less than useful error message on windows when privileges wrong for postgres

I have a client who mistakenly gave the postgres user on a windows  
machine admin privileges.

This mistake results in the service being unable to start up due to  
postgres refusing to start with admin privileges.

The error message from pg_ctl start -D bindir is PG_CTL.."could not  
locate matching postgres executable"

Is it not possible to provide a more useful error message ?

Dave

Dave Cramer <pg@fastcrypt.com> writes:
> I have a client who mistakenly gave the postgres user on a windows  
> machine admin privileges.

> This mistake results in the service being unable to start up due to  
> postgres refusing to start with admin privileges.

> The error message from pg_ctl start -D bindir is PG_CTL.."could not  
> locate matching postgres executable"

It's fairly hard to see how that mistake leads to that symptom.
Can you poke a bit more into exactly what is happening?
        regards, tom lane

On 21-Oct-08, at 5:24 PM, Tom Lane wrote:

> Dave Cramer <pg@fastcrypt.com> writes:
>> I have a client who mistakenly gave the postgres user on a windows
>> machine admin privileges.
>
>> This mistake results in the service being unable to start up due to
>> postgres refusing to start with admin privileges.
>
>> The error message from pg_ctl start -D bindir is PG_CTL.."could not
>> locate matching postgres executable"
>
> It's fairly hard to see how that mistake leads to that symptom.
> Can you poke a bit more into exactly what is happening?
>
find_my_exec returns -1 for a number of errors resulting in an error  
message much like above (wording is slightly different).

when I executed postgres.exe directly it complained of the user having  
admin privs

I removed admin privs from the pg user and pg_ctl worked fine.

I know it's not very specific, but fairly damning.

Dave

Dave Cramer <pg@fastcrypt.com> writes:
> On 21-Oct-08, at 5:24 PM, Tom Lane wrote:
>> It's fairly hard to see how that mistake leads to that symptom.
>> Can you poke a bit more into exactly what is happening?

> when I executed postgres.exe directly it complained of the user having  
> admin privs

With no arguments, that's not surprising; but pg_ctl invokes it with the
-V option, which should result in a version response occurring before
the root-privileges check (look at main/main.c).  So I think there's
something going on here that you've not identified.
        regards, tom lane