public inbox for [email protected]
help / color / mirror / Atom feedFrom: vrms <[email protected]>
To: [email protected]
Subject: Re: OS upgrade on postgres servers
Date: Tue, 17 Mar 2026 11:45:00 +0100
Message-ID: <[email protected]> (raw)
In-Reply-To: <CANzqJaBeJzmsF+OE8v8BR6v72WChr5F+4WUjRuzZ1dxb9UzwgA@mail.gmail.com>
References: <CAJk5AtbW=741yyD5+oV9yHSU40E4g57SY5DLRcJXsEK5-YFRig@mail.gmail.com>
<CANzqJaDdOPGTE3wjgVA4rsPsG2j44JjvDZLKw0DFnnwauEFw8g@mail.gmail.com>
<CAJk5AtaJkt1wETETV88LvY9NQwAC6KF35Q=L-rekGk=zLEfWcg@mail.gmail.com>
<CANzqJaBeJzmsF+OE8v8BR6v72WChr5F+4WUjRuzZ1dxb9UzwgA@mail.gmail.com>
@Ron: with 'parallel' you mean the -jX part of the pg_dump/pg_restore
commands you're suggesting, which would utilize multiple processes on
the server, right?
@rajeshkumar: It would also technically be possible to run a plain text
dump and pipe it through to psql on the fly. Something like
/usr/pgsql-17/bin/pg_dumpall -v -h <source-server> |
/usr/pgsql-17/bin/psql -h localhost |& tee /path/to/transfer.out
That way you have only one single action going on (on the target
server). You loose the option to interfere between dump and restore though.
You check the out file (/path/to/transfer.out) produced by this for any
errors (grep -iE 'warning|error|fatal|notice' /path/to/transfer.out)
afterwards.
Also you need to compare password_encryption on both ends and, if it
might be different (like md5 on the source, scram-sha-256 on the
target), set the passwords once again manually.
Assuming you have both servers running in parallel you can test those
options and see which one suits you while the source server is still
operating.
all best
Gunnar
On 3/16/26 21:00, Ron Johnson wrote:
>
>
> On Mon, Mar 16, 2026 at 3:45 PM Raj <[email protected]> wrote:
>
> Pgdg repo
> 100Gb
>
>
> That's relatively small. Parallel pg_dump/pg_restore should be pretty
> fast.
>
> Outage window - our decision. Client will accept our plan.
> Postgres upgrade may or may not be needed. Need help on both the
> scenarios
>
>
> What version of PG are you currently using? (Everything older than PG
> 14 is EOL, and PG 14 will go EOL this November.)
>
> I strongly recommend that you add the PGDG repository to yum/dnf, and
> then install the intended version (both before and after the upgrade
> to RHEL10.
>
> PG 17.latest or PG 18.latest are best, but of course you need to read
> the release notes and test your application against that new version.
>
> Then, for example:
> /usr/pgsql-17/bin/pg_dump -Fd -jX ....
> <upgrade RHEL to v10>
> /usr/pgsql-17/bin/pg_restore -Fd -jX
>
>
> On Tue, 17 Mar 2026, 00:41 Ron Johnson, <[email protected]>
> wrote:
>
> On Mon, Mar 16, 2026 at 2:57 PM Raj
> <[email protected]> wrote:
>
> Hi all,
>
> I have traditional servers with postgres with replication
> setup (primary - standby). OS team want to upgrade from
> rhel 8.10 to 10.
>
> As a dba, what is the suggestion we need to give. How do
> we proceed ? Should we stop the posygres servers? Should
> we get new servers with rhel 10 and migrate Data?
>
>
> That's certainly a safe method.
>
> What's the best procedure
>
>
> The main problem is collation change driven by the newer glibc
> version.
>
> 1. How did you install PG (from the RHEL repository, or from
> the PGDG repository)?
> 2. How big are your databases?
> 3. How big is your outage window?
> 4. Do you plan on upgrading Postgresql at the same time?
>
>
> --
> Death to <Redacted>, and butter sauce.
> Don't boil me, I'm still alive.
> <Redacted> lobster!
reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: [email protected]
Cc: [email protected], [email protected]
Subject: Re: OS upgrade on postgres servers
In-Reply-To: <[email protected]>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox