Re: Change pg_cancel_*() to ignore current backend

On 05/22/2015 03:35 PM, Andres Freund wrote:
> On 2015-05-22 17:29:03 -0500, Jim Nasby wrote:
>> On 5/22/15 4:51 PM, Tom Lane wrote:
>>> Jim Nasby <Jim(dot)Nasby(at)BlueTreble(dot)com> writes:
>>>> On 5/22/15 3:08 PM, Eric Ridge wrote:
>>>>> Rather than change the behavior of pg_cancel/terminate_backend(), why not change pg_stat_activity to exclude the current session? Seems like showing a row in pg_stat_activity for "SELECT * FROM pg_stat_activity" is kinda useless anyways.
>>>
>>>> Interesting idea. I suspect that would be even more invasive than
>>>> modifying the functions though...
>>>
>>> -1 ... some other columns in pg_stat_activity are potentially useful even
>>> for the current session, eg session and transaction start times.
>>
>> AFAICT the only field you can get in pg_stat_activity and nowhere else is
>> session start time (txn start is always now(), no?).
>>
>> If that's the only objection about eliminating the current backend from
>> pg_stat_activity then I'd say we should just add a session_start_time
>> function to handle that.
>
> That's far too big a backward compat break for something of very minor
> benefit.
>
> This whole discussion seems to be about making it easier to run SELECT
> pg_cancel_backend(pid) FROM pg_stat_activity;. But that shouldn't be
> made easier! If anything harder.

Linux doesn't prevent you from running Kill on your own process, either.
Would be interesting to find out how their discussion on the same topic
went.

--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com