Hi all,
I have a question about CleanupProc() and reaper().
CleanupProc() sends SIGUSR1 signal to all backends
if the exitStatus parameter is not 0.
Could we expect that other backends die with non
zero exit status ? Seems CheckPoint process could
be invoked after CleanupProc() is called and it
would remain alive unless SIGUSR1 is sent to the
process. The reset_shared() isn't called and so
the CheckPoint process may be blocked by spinlock/
BufferLock etc...
Regards.
Hiroshi Inoue