Hello,

I have a small network running windows.  Postgres 8.2.3 is the db and slony=
 version 1.2.9 is set up.

When I started replication my first db, I followed the examples and it work=
ed perfectly.  One master, one slave, db size of 300Mb.  Then I noticed tha=
t replication often fell behind.  If I watched the statistics info on pgAdm=
in III, it would report that the hanging event was a SYNC and often it woul=
d hang for several hours.  If I ignored the problem long enough, it would c=
atch up, run for a while, and then hang.  (Long enough was usually 12 - 24 =
hours).

My second db was larger, about 22 Gigs.  I tried for almost a month to repl=
icate it.  It never went through COPY_SET to completion.  I read everything=
 I could on line and tried every permutation of the slony.conf files.  I tr=
ied setting up with scripts and pgAdmin III.  It didn't matter.  Finally I =
remembered that wireless often drops the connection for a second or two at =
a time.  I can't rewire the office but did get one long cord, connect the w=
ireless router to one machine, and I was able to replicate the large db.

I still have the problem of both db's replication hanging for several hours=
.  I can only guess that slony is intolerant of even momentary network outa=
ges.  If I restart slony (net stop slony-i, net start slony-i) from the com=
mand line, it always catches up.

My question, is there any way of adjusting settings to make slony more tole=
rant?  Or am I looking in the wrong direction?  I know I could write anothe=
r service as a .NET program and implement ManagementEventWatcher to see whe=
n a connection resets and then restart slony-i that way, but that seems clu=
nky.

Any suggestions?


Thanks.


Al
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.slony.info/pipermail/slony1-general/attachments/20070516/=
2b004c47/attachment-0001.htm