Re: [HACKERS] What happens at BIND time? (pg_upgrade issue)

2013-08-30 Thread Josh Berkus
On 08/28/2013 11:44 AM, Josh Berkus wrote: > Tom, > >> Does the backend's memory usage climb, or hold steady? If the former, >> I'd bet on client failure to release resources, eg not closing the >> portals when done with them. A memory map from MemoryContextStats >> would help determine exactly

Re: [HACKERS] What happens at BIND time?

2013-08-28 Thread Josh Berkus
Tom, > Does the backend's memory usage climb, or hold steady? If the former, > I'd bet on client failure to release resources, eg not closing the > portals when done with them. A memory map from MemoryContextStats > would help determine exactly what's leaking. FS cache usage increases through t

Re: [HACKERS] What happens at BIND time?

2013-08-23 Thread Tom Lane
Josh Berkus writes: > So I'm looking for a cause for this. One thing would help is a clearer > understanding of what happens at BIND time for 3-phase queries. Planning. > Note that, in this workload, the user is re-parsing the queries > over-and-over rather than using prepared plans properly.

[HACKERS] What happens at BIND time?

2013-08-23 Thread Josh Berkus
All, Per -performance, we're trying to isolate and reproduce a customer issue where on 9.2 BINDs get progressively slower (as much as 2000% slower) over the course of a long test run, in a way they *don't* in 8.4. So I'm looking for a cause for this. One thing would help is a clearer understandi