I've recently (about a month now) been using Slony-1.1.0 to replicate 4
databases that I have.  3 are very small, but one is about 15G.

If left unchecked the slave server will run completely out of Swap memory in
about 24-36 hours.  The part I find odd is that physical memory never grows,
I would have expected that to run out first then swap.

I'm hoping that it's just a stupid user configuration error that I've made.
I can't seem to find anything in the archives that really fits the bill
here.

My Master and Slave servers are both Dual Xeon 3.06GHz Hyperthreaded
servers, with 1G physical memory, and 1G swap.  They are using 4 36G Ultra
320 drives arranged in 0+1 (2 drives combined in RAID 0 with the 2 sets as a
RAID 1).  The Master and Slave share a private 1Gb/s connection between
them, that is only used for Slony. These servers are identical hardware and
configuration, and are only used for our Postgres database.

The databases stay up to date and replicate well, the problem is that we
just run out of swap.  The Master does not show ANY growth of swap usage on
it.  There are no connection to the database on the Slave (but I would like
to eventually use it for select only queries).  I know this is a slony
related issue, because my "fix" right now, is to just stop/start slony on
the Slave and all the 
swap memory is freed.

My implementation is that I do have 4 completely separate databases so I'm
running 4 slon processes for basically 4 master/slave clusters. I'm
currently testing to see if running only one slon process demonstrates the
same problem, but I suspect it will.

Any help would be appreciated, let me know if you'd like further
information.  The 15G DB is pretty active, many selects and a reasonable
amount of INSERT and UPDATE.  It's our main Email repository for our users.

Thanks for the help.

--
David A. Niblett               | email: niblettda at gru.net
Network Administrator          | Phone: (352) 334-3400
Gainesville Regional Utilities | Web: http://www.gru.net/