Re: serverlog rotation/functions

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> Now that I understand Andreas's patch, and the way he is using shared
> memory to store only the timestamp, and how he checks shared memory on
> every elog call, I no longer see problems with his method.

The fundamentally unfixable problem with his method is that it can only
capture elog output, not stderr output from libraries that we don't
control (the dynamic linker being the biggest case, but there are
others).

I do not have any faith in the method as regards to switchover
reliability or message synchronization, either. I'm prepared to
grant that those might be only minor problems ... but I don't really see
why we should put up with doubts in this area. When you want to look at
a server log, it's because you are trying to debug a problem. The very
last thing you need is any niggling doubts about whether what you are
reading is the truth.

Finally, I simply do not trust *any* dependency on shared memory in this
connection. Again, I'm sure it works great in the normal case, but the
normal case is not what you need a server log for.

regards, tom lane