mastodon fails to startup

[robertbovens]

Name and Version

bitnami/mastodon App Version: 4.2.12 Package Version: 7.0.5

What architecture are you using?

amd64

What steps will reproduce the bug?

Mastodon runs well until I need to reboot the system or just restart kubernetes, mastodon won't be able to start. It only helps to redeploy mastodon-web container.

Are you using any custom parameters or values?

yes, but it does not have to do with a directory issue

What is the expected behavior?

Mastodon start without any problem

What do you see instead?

Mastodon won't start normally and the pod of mastodon-web gives the error:

Container: prepare-tmp-dir
Filter
Disconnected
mkdir: cannot create directory '/tmp/tmp-dir': File exists

[robertbovens]

By the way, it also helps to scale down mastodon-web to 0 and then to 1 again, then it starts normally

[dgomezleon]

Hi @robertbovens

Do you face the same problem with the latest version 9.1.0?

[robertbovens]

I am still on 4.2.12 Package Version: 7.0.5, i will try upgrade

[robertbovens]

I upgraded to Package Version: 8.0.0, same issue. I need more time to upgrade further

[robertbovens]

Upgrade to 9.0.4 didnt help either, i install with kubeapps and since kubeapps is giving me now sql error i cant upgrade, will try tomorrow.

[robertbovens]

Now I`m stuck when i want to upgrade further. I already reinstalled kubeapps, but when i try to upgrade to 9.0.5 or 9.1, kubeapps throws an error:
Unable to fetch the package "bitnami/mastodon (9.0.4)": Unable to get the available package detail for the package "bitnami/mastodon" using the plugin "helm.packages": internal: Unable to retrieve chart files: sql: no rows in result set.

Thanx

[migruiz4]

Hi @robertbovens,

That error is caused by an issue with the latest version of kubeapps.
We plan to release a new version of kubeapps (2.12.1) fixing this in the next days.

[robertbovens]

Ok thanx for the info @migruiz4

[migruiz4]

@robertbovens we have just released a new version of kubeapps (v2.12.1) which should fix the chart upgrade issue.

You can not install the bitnami/kubeapps chart version 17.1.1 that use the kubeapps version v2.12.1:
https://hub.docker.com/layers/bitnamicharts/kubeapps/17.1.1/images/sha256-a0bd2dcb9264e261e20a1f96dc6703e0fd3132524a991a085efac3364b8b7777

[robertbovens]

I reinstalled kubeapps, now i was able to upgrade mastodon. The problem is still there after rebooting/restarting kubernetes.
cannot create directory '/tmp/tmp-dir': File exists

When i scale down and scale up mastodon-web, everything resumes operations like it should be.

[dgomezleon]

Hi @robertbovens

Thanks for upgrading to the latest version.

By the way, it also helps to scale down mastodon-web to 0 and then to 1 again, then it starts normally

I was not able to reproduce the issue on GKE. This is what I tried:

1. helm install mastodon . -f ../../.vib/mastodon/runtime-parameters.yaml
2. Edit mastodon-web deployment to 0 replicas
3. Edit mastodon-web deployment to 1 replicas

It was properly started as you can see from the output below:

mastodon-apache-777fdf8f85-tl8gs        1/1     Running     0          34m
mastodon-elasticsearch-coordinating-0   1/1     Running     0          34m
mastodon-elasticsearch-data-0           1/1     Running     0          34m
mastodon-elasticsearch-ingest-0         1/1     Running     0          34m
mastodon-elasticsearch-master-0         1/1     Running     0          34m
mastodon-minio-697cd5f7fd-zq57l         1/1     Running     0          34m
mastodon-minio-provisioning-zmzcw       0/1     Completed   0          34m
mastodon-postgresql-0                   1/1     Running     0          34m
mastodon-redis-master-0                 1/1     Running     0          34m
mastodon-sidekiq-6d8b5cdcd5-xjlmg       1/1     Running     0          34m
mastodon-streaming-75d7755976-glmqg     1/1     Running     0          34m
mastodon-web-7d769f6d88-kmz22           1/1     Running     0          55s

Could you please le me know what could differ from your scenario?

[robertbovens]

My scenario is:
no apache, no elasticsearch and no minio, i use pvs.
I have all resourcesPreset set to none.

I took my settings to other vm with kubernetes. Installed with helm3.
The install is fine, and it works. The problem is ONLY and ONLY when i reboot the vm or restart kubernetes, then the error still exist: I even put all settings to standard settings, so now it boots also elasticsearch, apache etc, but i also end with the same problem AFTER a reboot of the vm:

CrashLoopBackOff (back-off 5m0s restarting failed container=prepare-tmp-dir pod=mastodon-web-6ffc96ffbc-p2jgk_mastodon-test(672133b6-062e-4630-bd4f-5ffefab4ad70)) | Last state: Terminated with 1: Error, started: Thu, Dec 26 2024 2:51:10 pm, finished: Thu, Dec 26 2024 2:51:11 pm

mkdir: cannot create directory '/tmp/tmp-dir': File exists

At this moment i need to set mastodon-web to 0 deployments and then 1, then it boots like it should. I need to tell you that i use rancher for my kubernetes vms, so i have a + and a - button to scale my deployment.

Thanks again for your support

[robertbovens]

• mastodon-elasticsearch-coordinating-0 1/1 Running 1 (11m ago) 26m
• mastodon-elasticsearch-data-0 1/1 Running 1 (11m ago) 26m
• mastodon-elasticsearch-ingest-0 1/1 Running 1 (11m ago) 26m
• mastodon-elasticsearch-master-0 1/1 Running 1 (11m ago) 26m
• mastodon-init-jr2dm 0/2 Completed 0 26m
• mastodon-postgresql-0 1/1 Running 1 (11m ago) 26m
• mastodon-redis-master-0 1/1 Running 1 (11m ago) 26m
• mastodon-sidekiq-84cbbc7d48-g8tfd 0/1 Unknown 0 26m
• mastodon-streaming-6d8948d45c-dzmkn 0/1 Unknown 0 26m
• mastodon-web-6ffc96ffbc-p2jgk 0/1 Unknown 5 26m

[robertbovens]

I can see now a difference, my init job is still in the list, not sure if it tells you anything, but yours doesn`t show that. Not sure if it has something to do with it or not though.