Re: function for testing that causes the backend to terminate

2021-05-03 Thread Dave Cramer
Joe,

Thanks,

This works and I don't have to install anything!

Dave Cramer


On Thu, 29 Apr 2021 at 16:16, Joe Conway  wrote:

> On 4/29/21 6:56 AM, Dave Cramer wrote:
> > For testing unusual situations I'd like to be able to cause a backend to
> > terminate due to something like a segfault. Do we currently have this in
> > testing ?
>
> If you can run SQL as a superuser from that backend, try:
>
> COPY (SELECT pg_backend_pid())
>   TO PROGRAM 'xargs kill -SIGSEGV';
>
> HTH,
>
> Joe
>
> --
> Crunchy Data - http://crunchydata.com
> PostgreSQL Support for Secure Enterprises
> Consulting, Training, & Open Source Development
>


Re: function for testing that causes the backend to terminate

2021-05-01 Thread Andrew Dunstan


On 4/29/21 4:16 PM, Joe Conway wrote:
> On 4/29/21 6:56 AM, Dave Cramer wrote:
>> For testing unusual situations I'd like to be able to cause a backend
>> to terminate due to something like a segfault. Do we currently have
>> this in testing ?
>
> If you can run SQL as a superuser from that backend, try:
>
> COPY (SELECT pg_backend_pid())
>  TO PROGRAM 'xargs kill -SIGSEGV';
>
> HTH,
>
> Joe
>


or a plperlu function that does 'kill 11, $$;' should do it.


cheers


andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com





Re: function for testing that causes the backend to terminate

2021-04-30 Thread Bharath Rupireddy
On Thu, Apr 29, 2021 at 4:36 PM Bharath Rupireddy
 wrote:
>
> On Thu, Apr 29, 2021 at 4:27 PM Dave Cramer  wrote:
> > For testing unusual situations I'd like to be able to cause a backend to 
> > terminate due to something like a segfault. Do we currently have this in 
> > testing ?
>
> Well, you could use pg_terminate_backend which sends SIGTERM to the
> backend. However, we don't have a function that sends SIGSEGV yet, you
> could signal the backend with SIGSEGV directly, if possible.

And, I came across an extension called pg_crash [1], see if that helps.

[1] 
https://www.cybertec-postgresql.com/en/pg_crash-crashing-postgresql-automatically/

With Regards,
Bharath Rupireddy.
EnterpriseDB: http://www.enterprisedb.com




Re: function for testing that causes the backend to terminate

2021-04-29 Thread Joe Conway

On 4/29/21 6:56 AM, Dave Cramer wrote:
For testing unusual situations I'd like to be able to cause a backend to 
terminate due to something like a segfault. Do we currently have this in 
testing ?


If you can run SQL as a superuser from that backend, try:

COPY (SELECT pg_backend_pid())
 TO PROGRAM 'xargs kill -SIGSEGV';

HTH,

Joe

--
Crunchy Data - http://crunchydata.com
PostgreSQL Support for Secure Enterprises
Consulting, Training, & Open Source Development




Re: function for testing that causes the backend to terminate

2021-04-29 Thread Bharath Rupireddy
On Thu, Apr 29, 2021 at 4:27 PM Dave Cramer  wrote:
> For testing unusual situations I'd like to be able to cause a backend to 
> terminate due to something like a segfault. Do we currently have this in 
> testing ?

Well, you could use pg_terminate_backend which sends SIGTERM to the
backend. However, we don't have a function that sends SIGSEGV yet, you
could signal the backend with SIGSEGV directly, if possible.

With Regards,
Bharath Rupireddy.
EnterpriseDB: http://www.enterprisedb.com




function for testing that causes the backend to terminate

2021-04-29 Thread Dave Cramer
For testing unusual situations I'd like to be able to cause a backend to
terminate due to something like a segfault. Do we currently have this in
testing ?

Dave Cramer