Shaun Thomas wrote:
> Hey guys,
>
> I think there's a major problem with the 1.2.1 branch.  So far, it
> seems to be coming from slonik_init_cluster.  I started with a new
> cluster that (almost) worked with 1.2.0, and to make sure it was fresh,
> I blew away everything.  slonik_uninstall_nodes, the works.
>
> The new cluster I set up failed pretty much immediately.  I ran the
> script slonik_init_cluster generated and node 3 was only entered on
> nodes 1 and 3, no paths were stored, and every time I tried to subscribe
> a set, the initial table copy never started.  I fixed this by running
> slonik_store_node manually for each node, and cutting/pasting the paths
> into a file and running it against slony.  Then of course node 3 wasn't
> working (node 2 had no record it existed), so I uninstalled it, and used
> slonik_store_node to rebuild it.
>
> What on earth is going on here?  Missing node entry?  No paths?  I'm
> glad the log switching code is fixed in 1.2.1, but if the initial setup
> script doesn't work, I'm a little leery about what else might be wonky
> in there.
>   
Look, the last developer doing serious work on the "altperl" scripts had
a change in involvements about a year and a half ago, and has ceased his
efforts.

I did some (evidently somewhat broken) remediation in 1.2; I committed
some changes this week that I believe resolve your issue.

But you need to be aware that these scripts currently do not have any
"attentive developer," and also that none of the 'official' testing of
the engine involves those scripts.  They're not getting much in the way
of "tender care."  That suggests that there is a bit of a problem with
trusting those scripts terribly far.