Re: rebellious pg stats collector (reopened case)

Tom Lane wrote:
> Alvaro Herrera <alvherre@commandprompt.com> writes:
>
>> Tom Lane wrote:
>>
>>> I wonder whether your tracing tool is affecting the result of
>>> getppid().  Most people would consider that a bug in the tracing tool.
>>>
>
>
I wrote to an official the FreeBSD list about this getppid() problem but
got no answer other than that "this behaviour is documented". :-(

The problem is still there:


  PID USERNAME       THR PRI NICE   SIZE    RES STATE  C   TIME   WCPU
COMMAND
11205 pgsql            1 104    0 22400K  7112K CPU5   5 159.7H 99.02%
postgres


100% CPU since 159 hours! What can I do? Instead of tracing system
calls, is there a way to start the stats collector in debug mode? Or
maybe is it possible to change the source code, and disable the "is
postmaster alive" check for testing?

Thanks