Re: Parallel pg_dump's error reporting doesn't work worth squat

At Thu, 2 Jun 2016 12:17:11 -0400, Alvaro Herrera <alvherre@2ndquadrant.com> wrote in
<20160602161711.GA239156@alvherre.pgsql>
> Tom Lane wrote:
> > Kyotaro HORIGUCHI <horiguchi.kyotaro@lab.ntt.co.jp> writes:
> > > Apart from the invalid snapshot problem, I looked the patch
> > > previously mentioned mainly for Windows.
> > 
> > Thanks for looking!
> > 
> > > Even though the threads started by beginthread cannot be
> > > terminated cleanly from outside, but the whole process will soon
> > > terminate anyway, so we could use TreminateThread. This seems
> > > working. (Attached patch)
> > 
> > Seems reasonable to me; I was unhappy about the lack of any direct
> > equivalent to the child SIGTERMs that the Unix code does.

For sure, any of the "dangers" of TerminateThread don't matter
for this case.

https://msdn.microsoft.com/en-us/em-us/library/windows/desktop/ms686717(v=vs.85).aspx

> If the target thread owns a critical section, the critical
> section will not be released.
> 
> If the target thread is allocating memory from the heap, the heap
> lock will not be released.
> 
> If the target thread is executing certain kernel32 calls when it
> is terminated, the kernel32 state for the thread's process could
> be inconsistent.
> 
> If the target thread is manipulating the global state of a shared
> DLL, the state of the DLL could be destroyed, affecting other
> users of the DLL.



> Given this testing, it's clear that the timeout on select() is useless;
> we could get rid of it in vacuumdb.c too.  I'll post a patch later.

Agreed. Command pipes may be in blocking mode for the case.

regards,

-- 
Kyotaro Horiguchi
NTT Open Source Software Center