Issues with upgrading from 1.7.0 to 2.1.0

General Discussions
21

Inside the .mender file without migration I still have the “ArtifactCommit_Enter_10_migrate-configuration” script

22

Yeah, that is the problem and why it fails.

Unclear why it would be still there though. Might be a sstate cache problem again

23

Yes it could be a sstate cache problem. But I’m not so sure, because “bitbake -c clean mender && bitbake -c cleansstate mender && bitbake -c my-image && bitbake -c cleansstate my-image” should fix the problem, but not in this case.

The migration script is always included in the mender file, also with everything disabled in local.conf.

Do you have suggestions to fix this issue? Is there a trick?

PS: If remove quickly the script from the mender file, can I try to install the update without break something? Only to be sure that everything is ok, except for the script problem, at least I want to understand If I can upgrade successfully with this configuration.

24

Try removing this folder,

${DEPLOY_DIR_IMAGE}/mender-state-scripts

Looking through the code there is nothing that would clean this up and the Mender Artifact bbclass will include scripts from this folder if they exist

25

Removing that folder, It worked perfectly. You are right, that folder was always there. It should be removed or ignored when unused.

So I successfully migrated from 1.7.0 to 2.1.0 and then to 2.1.0 to 2.1.0, removing that folder.

Thanks for the support
:+1:

1 Like
26

Thank you for reporting back all the way and you testing this flow is very valuable to us.

We will need to take a look at how we can improve the documentation on this, making the migration easier.

1 Like
27

Update in progress,

28

Hey guys,

Has documentation outlining the 1.7 -> 2.1 upgrade been published yet? If so, where?

Thanks!
SLR-

29

Bump.

We have a production release scheduled for the last week of November that was expected to include a client update. Based on what I am reading here, there isn’t yet an official path from 1.7 to 2.1, at least not documented. Is such a document imminent? I’m trying to gauge whether we need to pivot to mitigate the apparent risk in this upgrade.

SLR-

30

Sorry for the delay. It is currently documented here,

https://docs.mender.io/2.2/troubleshooting/running-yocto-project-image#upgrading-from-thud-to-warrior-fails-with-dual-rootfs-configurat

Even though it is under the 2.2 version of release docs, I believe is is applicable for 2.1 as well.

Also not that we are releasing 2.2 at the beginning of next week.

1 Like