Cloudron Forum

I decided to take some effort upon myself to see if I could package PhotoPrism. I've got a version I've uploaded to Cloudron gitlab that seems good enough to use right now. Some notes:

• The software expects either MariaDB or Sqlite. I tried hooking up to MySql, but there were syntax issues. The current config uses sqlite, but can be changed to PGSQL once they implement that into PhotoPrism.
• PhotoPrism needed a newer go than was in the base cloudron image, so I found it easiest to remove the old go version and inplace put the newer version. This can be removed once base image is updated.
• To make things easier, I did use the precompiled tensorflow from PhotoPrism instead of compiling myself. The version I included requires AVX2 (https://en.wikipedia.org/wiki/Advanced_Vector_Extensions#CPUs_with_AVX2)
• There might be a better way to handle the env vars than I did, but it works for now.

Hopefully this can be useful to someone wanting this software now or to help make something more official.

https://git.cloudron.io/eyecreate/photoprism

I was just working on a package for this too. I too noticed that there were areas that were hard coded to the docker host names. There is this spot where the microservices want to access the machine-learning part that could use a patch like you made for the api server. (as well as supporting the machine-learning service too)
https://github.com/immich-app/immich/blob/1ba998aa68ad91aa3951c81fdf3f0570b7e2b704/server/apps/microservices/src/processors/metadata-extraction.processor.ts#L129

I took some time tonight and made a first version of the software package. I'm still testing it out, but if you have been wanting to try out Kavita, you can try it out here: https://git.cloudron.io/eyecreate/kavita
(use the build instructions)

An update a bit ago to the cloudron file browser made it harder to use on mobile. File names don't show up unless rotated to landscape and space is taken up on right hand side with larger file details that could be used to browse better. I would like it if the experience was improved for mobile, especially in portrait rotation.

I was playing around with homeassistant to see of I could use this server app instead of my RPI. It seems I can't find a way to pass in devices like my USB ZigBee radio to the container. Is there a workaround or plan to make this possible?

I don't have any ipv6 only clients, but I really would like to support the sites on my cloudron being accessed over ipv6 seamlessly.

I noticed recently that my subscription list of videos was not updating. After looking into the documentation, it seems that they expect you to reboot the software frequently. Anywhere from hourly to daily. After rebooting, it was working correctly again. Is there a way that we could automate or allow configuring how often a container is rebooted?

@murgero haven't tested extensively, but cloudron supports sftp access for all containers now.

@girish I saw something similar in my logs for Storj. I contacted support and they mentioned it's on thier roadmap for the next few months.

@girish I have changed the update schedule and will get back with you if I notice nothing has changed.

Yep, just added it now!

I've also had similar issues with backups the past few days.

2024-03-20T08:58:00.790Z box:shell backup-snapshot/app_c4a27e93-af6e-44e2-b7cf-9056fa01c5c3 code: null, signal: SIGKILL
2024-03-20T08:58:00.792Z box:taskworker Task took 10678.952 seconds
2024-03-20T08:58:00.793Z box:tasks setCompleted - 16169: {"result":null,"error":{"stack":"BoxError: Backuptask crashed\n    at runBackupUpload (/home/yellowtent/box/src/backuptask.js:164:15)\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n    at async uploadAppSnapshot (/home/yellowtent/box/src/backuptask.js:361:5)\n    at async backupAppWithTag (/home/yellowtent/box/src/backuptask.js:383:5)\n    at async fullBackup (/home/yellowtent/box/src/backuptask.js:504:29)","name":"BoxError","reason":"Internal Error","details":{},"message":"Backuptask crashed"}}
2024-03-20T08:58:00.793Z box:tasks update 16169: {"percent":100,"result":null,"error":{"stack":"BoxError: Backuptask crashed\n    at runBackupUpload (/home/yellowtent/box/src/backuptask.js:164:15)\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n    at async uploadAppSnapshot (/home/yellowtent/box/src/backuptask.js:361:5)\n    at async backupAppWithTag (/home/yellowtent/box/src/backuptask.js:383:5)\n    at async fullBackup (/home/yellowtent/box/src/backuptask.js:504:29)","name":"BoxError","reason":"Internal Error","details":{},"message":"Backuptask crashed"}}
BoxError: Backuptask crashed

It would be nice if the error or code it pointed too had more info about what happened. Best I can tell is it got killed somehow. I increased the backup task memory and will see what happens next time.

Edit: it didn't crash yesterday, so maybe it just needed more ram.

Yes, I am interested in moving immich to oidc without reinstalling.

I've made a quick cloudron build here as a merge of River and Emby repos.
https://git.cloudron.io/eyecreate/jellyfin-app

@nebulon Adding memory didn't do anything, but taking your direction to erase database ended up working out.

Note that /var/lib/.initialized didn't exist and was in the graphite directory. Also, I had to manually stop and start the container and execute the command right away or docker wouldn't let me do anything because the container was in the process of restarting/starting.

@girish I found out through some testing that oidc login on android worked if Firefox was the default(custom tab) browser but not if Chrome was.

I haven't updated the readme(which references some incorrect instructions for install), but I think the build instructions should work.

It seems that switching the backup format from rsync to tgz has allowed the backup to complete again.

For a few months, I've had issue with all my apps failing to auto-update. I usually get a message like this:

Docker Error: (HTTP code 404) no such container - No such image: cloudron/com.gitlab.cloudronapp:20220504-210413-802d48119

If I go to the repair tab, I can manually trigger it to succeed on each step of update and configure, but it seems to be unable to succeed itself.

Any idea on why this is happening?

Now with Docker this year mentioning no long term storage of docker images for free accounts, this will be even more valuable to have in gitlab.