deployment on master fails

[markus2330]

Describe the Bug

https://build.libelektra.org/blue/organizations/jenkins/PermaPlanT-Folder%2FPermaplanT/detail/master/304/pipeline

Steps to Reproduce the Problem

trigger master build

Expected Result

all green 🟢

Actual Result

+ sudo /usr/local/bin/permaplant-deploy.sh dev /home/jenkins/workspace/maPlanT-Folder_PermaplanT_master
+ INSTANCE_PATH=/opt/permaplant/dev
+ BACKEND_BINARY=/home/jenkins/workspace/maPlanT-Folder_PermaplanT_master/backend/target/release/backend
+ SERVICE=permaplant-dev
+ [[ ! -d /home/jenkins/workspace/maPlanT-Folder_PermaplanT_master ]]
+ [[ ! -d /opt/permaplant/dev ]]
+ cd /home/jenkins/workspace/maPlanT-Folder_PermaplanT_master/backend
+ LC_ALL=C
+ diesel --version
diesel 2.0.1
+ cd /home/jenkins/workspace/maPlanT-Folder_PermaplanT_master/backend
+ LC_ALL=C
+ diesel migration run --locked-schema
Running migration 2023-07-03-165000_heatmap
Failed to run 2023-07-03-165000_heatmap with: column "geometry" of relation "maps" contains null values
script returned exit code 1

PermaplanT version

master

Operating System

Linux

Browser

Firefox

Versions of other relevant software/extensions

No response

Additional info/logs

No response

[4ydan]

I have seen this error in a PR that was probaly merged now.

[4ydan]

This PR #622
Hm idk.
https://build.libelektra.org/job/PermaPlanT-Folder/job/PermaplanT/job/release%252F0.2.3/1/display/redirect
https://build.libelektra.org/job/PermaPlanT-Folder/job/PermaplanT/job/PR-622/1/display/redirect

[markus2330]

Thx! Can you also add the link to the build log?

Btw. the PR must have some changes to migrations...

[4ydan]

Yes, I think we should really not merge stuff that doesnt go completely through the pipeline.
Something like this is really dangerous

Just because only mdbook fails, doesnt mean everything else is fine, we need to wait for the deploy stage. It's harder to track down now.

[markus2330]

I agree.

My best guess it that the faulty one is: #595

We don't have yet the #134, so it slipped through (even though being successfully build).

If there are no other ideas, I would revert #595

[4ydan]

I can try some commits out. Up to you.

[4ydan]

I will checkout #595 and test it locally.

[4ydan]

Yes its that branch.
Interesting though, because the branch was rebased, right? And all checks were passing?

[4ydan]

Hm, I cant reproduce it anymore, what should be done after resetting the database to reproduce it?

[markus2330]

You cannot reproduce it anymore after your have reset the database, this is in the nature of these migrations.

I created #666 to revert. Unfortunately, it cannot be tested on pr.libelektra.org, so I need to merge to master to see if it works.

[markus2330]

Probably the better option is to merge #644 and reset the databases already for this release (and hope we can stay compatible). Let us see if it builds.

[4ydan]

Also should we maybe rethrow the exception to make the pipeline fail or do we leave it like this?

[markus2330]

Absolutely! I just noticed that https://build.libelektra.org/blue/organizations/jenkins/PermaPlanT-Folder%2FPermaplanT/detail/master/303/pipeline also already failed.

As #644 is not ready yet we will probably need to do the resetting of DBs twice.

I already resetted dev.permaplant.net, let us see if https://build.libelektra.org/blue/organizations/jenkins/PermaPlanT-Folder%2FPermaplanT/detail/master/305/pipeline will work.

[markus2330]

Worked! Now release: https://build.libelektra.org/job/PermaPlanT-Folder/job/PermaplanT-Release/14/ (also resetted the DB)

[markus2330]

Also worked, 0.2.3 is now released!