Ok, I figured out why the Makefile is broken for pglogical for PostgreSQL V.10. The pg_config utility returned 10.4 for the line setting Makefile parameter PGVER. Adding a simple “| cut -c1-2 to that line allows the Makefile to run for 9.6 and 10.4, probably other versions as well. There are compile problems, however, in several source files. I don’t have the code in front of me, but one doing heap work calls replorigin_drop with a single argument, but that changed in V10 to need 2 args. Fairly easy to fix that one. Another source calls ExecARUpdateTriggers, and its friends Insert and Delete, those signatures have changed as well, and I haven’t yet figured out how to alter pglogical for compiling these. If anyone reading here is from 2nd Quadrant,I’m curious if this is actually being maintained. I did post these as an issue on github. — Jay Sent from my iPad > On Jun 26, 2018, at 6:47 PM, David Morton <david.morton@xxxxxxxxx> wrote: > > Take a look at pglogical ... we've used it for 9.4 -> 9.6 and the same theories apply going to 10.x I imagine. People have used it in combination with pgbouncer to achieve near outage-less upgrades. Some helpful links below: > https://blog.2ndquadrant.com/pglogical-logical-replication-postgresql-10/ > https://blog.2ndquadrant.com/near-zero-downtime-automated-upgrades-postgresql-clusters-cloud/ > https://blog.2ndquadrant.com/near-zero-downtime-automated-upgrades-postgresql-clusters-cloud-part-ii/ > > > On 27/06/18, 10:38 AM, "John Scalia" <jayknowsunix@xxxxxxxxx> wrote: > > Hi all, > > My employer is in the process of upgrading PostgreSQL, and they would like me to make this happen without any real interruption of service. As I understand it, 9.6 does not support the publish/subscribe model like 10.4 does, but logical replication is the basis of that model. So, does anyone have any suggestions for how to implement the receiver side on a 10.4 server? Or could you at least point me to some sort of how-to page? > > TIA, > Jay > > > Sent from my iPad > >
