We have a few production sites running ES 7.16 and would like to upgrade to 8.6:
- upgrade will be done with full cluster restart
- all indices were created under ES 7.x
- we have checked Upgrade Assistant on ES 7.16 and confirmed all is ok
However, we noticed that there is a requirement (Upgrade to Elastic 8.6.2 | Elastic Installation and Upgrade Guide [8.6] | Elastic) to perform multi upgrades i.e. upgrade to 7.17 then 8.6. Is there any reason why we need to do this? We would like to minimise the deployment steps.
7.16 was released some time before 8.0 so its upgrade assistant does not cover everything, nor does it have all the needed deprecation warnings. The only way to verify for sure that you will encounter no problems in the upgrade is to use 7.17 first.
Thanks, that make sense. Apart from the deprecation warning, is there anything else that requires us to upgrade to 7.17 first?
All of our deployments are identical. So, we are planning to "simulate" the upgrade on our test environment, which has the identical configuration & settings, by upgrading it to 7.17 first then to 8.6. Based on our findings in the test environment, then we can apply the same migration issue fix to the production environment. Basically, to avoid having to upgrade to 7.17 in the production environment.
I'm not aware of anything else, but that doesn't mean it's safe to skip the upgrade to 7.17. The process described in the manual is the recommended one, and the one covered by tests. If it were me and my production data then I'd avoid taking risks like this.
Just double checking, if we have indices created on ES 6.x, do we have to reindex them on ES 7.17? Or can we do it in ES 7.16 or earlier?
Any 7.x version is fine, including 7.16.x and 7.17.x
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.