Re: PATCH: backtraces for error messages

On 2018-Jun-20, Craig Ringer wrote:

> Hi folks
>
> I recently needed a way to get backtraces from errors in a convenient,
> non-interactive and indescriminate way. The attached patch is the result.
> It teaches Pg to use libunwind to self-backtrace in elog/ereport.
>
> Anyone think this is useful/interesting/worth pursuing?

I think we sorely need some mechanism to optionally get backtraces in
error messages. I think having backtraces in all messages is definitely
not a good thing, but requiring an explicit marker (such as in my patch)
means the code has to be recompiled, which is not a solution in
production systems. I kind lean towards your approach, but it has to be
something that's easily enabled/disabled at runtime.

I have no idea how expensive backtrace() and libunwind are, but I doubt
we want to pay the cost for every message before we know that error
requires the backtrace to be collected. Something like PGC_USERSET
server_min_backtraces=PANIC
might be a possible interface.

--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services