Thread: Postgresql community edition upgrade
Hi all,
Can we upgrade the postgresql community edition version from 11 to 14 directly??
Am 12.10.22 um 15:43 schrieb Ebin Jozer: > Hi all, > Can we upgrade the postgresql community edition version from 11 to 14 > directly?? Yes, all versions from 8.4 can be upgraded to the latest version directly. -- Holger Jakobs, Bergisch Gladbach, Tel. +49-178-9759012
Attachment
Thanks holger.
How to find it??
And
What would be the rollback step in case the upgrade failed while using pg_upgrade??
On Wed, 12 Oct, 2022, 7:24 pm Holger Jakobs, <holger@jakobs.com> wrote:
Am 12.10.22 um 15:43 schrieb Ebin Jozer:
> Hi all,
> Can we upgrade the postgresql community edition version from 11 to 14
> directly??
Yes, all versions from 8.4 can be upgraded to the latest version directly.
--
Holger Jakobs, Bergisch Gladbach, Tel. +49-178-9759012
> On Oct 12, 2022, at 11:18 AM, Ebin Jozer <ebinjozer@gmail.com> wrote: > > What would be the rollback step in case the upgrade failed while using pg_upgrade?? - pg_upgrade does pretty extensive pre-flight checks before proceeding - it does not modify original files, but copies (& modifies the copied catalog files as needed) (I think even if you use link mode, it doesn't modify original catalog files, so you can start the old one until you havestarted the new one. I'd look for more info from someone more knowledgeable about this exact process.) I've personally never seen a problem once the pre-flight passes and it proceeds.
On Wed, 2022-10-12 at 22:48 +0530, Ebin Jozer wrote: > What would be the rollback step in case the upgrade failed while using pg_upgrade?? If you don't use --link, the old cluster will still be there and can be used. If you use --link, restore your backup. Yours, Laurenz Albe -- Cybertec | https://www.cybertec-postgresql.com
On 12 October 2022 18:29:12 Scott Ribe <scott_ribe@elevated-dev.com> wrote:
(I think even if you use link mode, it doesn't modify original catalog files, so you can start the old one until you have started the new one. I'd look for more info from someone more knowledgeable about this exact process.)
Correct. I've had it go wrong with link mode. As long as you don't start the new cluster, the old one still works absolutely fine with a minor (documented) step.
IIRC, the data files (in /data/base) are linked (and unchanged by the process). The control files, transaction logs, etc aren't, but the old control files are "disabled" to prevent accidental starting of the old cluster (it's easy to re-enable it)
See (17) of the Usage notes
Paul
--
Paul Smith Computer Services
Tel: 01484 855800
Vat No: GB 685 6987 53
Hi thanks for the response so far.
We have master and slave environment. So for upgrade we have upgrade first in slave and then in master environment. Please confirm
On Thu, 13 Oct, 2022, 2:13 am Paul Smith, <paul@pscs.co.uk> wrote:
On 12 October 2022 18:29:12 Scott Ribe <scott_ribe@elevated-dev.com> wrote:(I think even if you use link mode, it doesn't modify original catalog files, so you can start the old one until you have started the new one. I'd look for more info from someone more knowledgeable about this exact process.)Correct. I've had it go wrong with link mode. As long as you don't start the new cluster, the old one still works absolutely fine with a minor (documented) step.IIRC, the data files (in /data/base) are linked (and unchanged by the process). The control files, transaction logs, etc aren't, but the old control files are "disabled" to prevent accidental starting of the old cluster (it's easy to re-enable it)See (17) of the Usage notesPaul--Paul Smith Computer ServicesTel: 01484 855800Vat No: GB 685 6987 53
> On Oct 13, 2022, at 3:42 AM, Ebin Jozer <ebinjozer@gmail.com> wrote: > > We have master and slave environment. So for upgrade we have upgrade first in slave and then in master environment. Pleaseconfirm Depends on your HA requirements. Easiest: shut down primary and all db access, upgrade one, bring it back up as primary, upgrade other and bring it back onlineas replica. Harder: use a replication scheme that works across versions (publish/subscribe, pglogical, londiste, bucardo) so that youcan bring up a new replica on the new version, switch over to it with minimal downtime (just enough to make sure changesfrom master have been replicated), then turn the former primary into a new replica.