On 2014-07-29 1:07 AM, Chris Lu wrote:
> Thanks for the info! I do found out that RedShift crashed.
>
> However, I am not really doing anything special except sending 2 queries,
> one by one. How come pq crashes the RedShift? Do you have any clue?
I don't remember the exact conditions under which it occurs, and I
stopped being interested once Amazon refused to look at the minimal Go
program which reproduced the problem. What pq does is completely valid
and postgres 8.0.2 (which they claim to have branched off of) doesn't
crash given the same exact sequence of messages -- at least in the test
case I had the last time.
Perhaps if you care enough, you could construct a program in their
language of choice which simply uses TCP sockets to send the correct
sequence of bytes to log in and crash the server.
.marko