Re: Help in vetting my steps for Postgres DB upgrade from Ver 13.X to ver 15.X

Team
I followed Greg suggested steps .One of big had only one table and around four million recordsi am doing dev env restoration into new vmthe target VM env is an POC server and took 3 hrs to restore four million records.Now it is doing process of lo_open / lo_close /  lowrite  etci.e pg-dump-creates-a-lot-of-pg-catalog-statements
is there any alternate way , to speedup  this process.
i can see in the select count(*) record count is matching (target and source)
Regards

On Wednesday, December 4, 2024 at 10:47:26 AM EST, Greg Sabino Mullane <htamfids(at)gmail(dot)com> wrote:

On Wed, Dec 4, 2024 at 7:42 AM Bharani SV-forum <esteembsv-forum(at)yahoo(dot)com> wrote:
a) is the above said steps is correct with the given existing and proposed setup

No. Here are some steps:
* Install Postgres on the new VMHowever you get it, use the newest version you can. As of this writing, it is Postgres 17.2. Version 15 is okay, but going to 17 now means a better Postgres today, and no worrying about replacing v15 in three years.
* Create a new Postgres clusterOn the new VM, use the initdb command to create a new data directory.Use the --data-checksums option
* Start it upAdjust your postgresql.conf as neededAdjust your pg_hba.conf as neededInstall any extensions used on the old VMStart the cluster using the pg_ctl command (or systemctl)
* Test connection to the old vm from the new vmOn the new vm, see if you can connect to the old one:psql -h oldvm -p 5432 --listYou may need to adjust firewalls and pg_hba.conf on the old vm.
* Copy the dataRun this on the new VM, adjusting ports as needed:time pg_dumpall -h oldvm -p 5432 | psql -p 5432
Bonus points for doing this via screen/tmux to prevent interruptions
* Generate new statistics and vacuumOn the new vm, run:psql -c 'vacuum freeze'psql -c 'analyze'
* Test your application
* Setup all the other stuff (systemd integration, logrotate, cronjobs, etc.) as needed
As Peter mentioned earlier, this can be done without disrupting anything, and is easy to test and debug. The exact steps may vary a little, as I'm not familiar with how Amazon Linux packages Postgres, but the basics are the same.
Take it slow. Go through each of these steps one by one. If you get stuck or run into an issue, stop and solve it, reaching out to this list as necessary.
Cheers,Greg