Robert Haas <robertmh...@gmail.com> writes: > On Tue, Sep 11, 2018 at 5:54 AM, Simon Riggs <si...@2ndquadrant.com> wrote: >> Please explain why you think that would be with no restart.
> Because the startup process will die, and if that happens, IIRC, > there's no crash-and-restart loop. You're just done. Unless we think that the startup process will never never ever throw an error, that might be a behavior that needs discussion in itself. Obviously an infinite crash-and-restart loop would be bad, but perhaps the postmaster could have logic that would allow restarting the startup process some small number of times. I think the hard part would be in deciding whether a previous restart had succeeded (ie made progress beyond the prior crash point), so that it should no longer count against the retry limit. regards, tom lane