Allow using a pre-downloaded index file

[kevinior]

I've been having some problems with downloading the index file in Docker, for some reason it keeps ending up corrupted after downloading for over an hour each time.

To work around this I've downloaded the file on the Docker host and manually unpacked it, but that's quite a fiddly process.

It would be nice if photon-docker could be configured check in some location for an already-downloaded index file and it could take care of the unpacking. That way, if I have a known-good index file I can always fall back to it if the automatic update fails.

[eng3]

Yeah, like check a specific path that we can bind mount and use that instead of downloading.
Probably a pretty simple change to start_photon.sh

[SmokeShine]

Bumping this up. @rtuszik is it possible to push this as a small bug fix instead of a full release?

Issue Description:

Downloaded pbf file for France and mounted to a folder. Docker start downloading the completing 110 Gb file. I assume that the start_proton.sh sees the photon data folder is empty and starts downloading the complete file.

[rtuszik]

I have to admit that I don't quite understand the point of this feature suggestion. If you are already downloading the index yourself, I would suggest that you do so with the provided commands, e.g.: wget -O - https://download1.graphhopper.com/public/photon-db-latest.tar.bz2 | pbzip2 -cd | tar x. This will extract straight away leaving you with the photon-data directory that you can mount to the container.

@SmokeShine In your case the issue is with the fact that you have a PBF file. Photon does not work with PBF files.

[kevinior]

For some reason I thought it was more complicated than that to unpack the index into the right directories with the right permissions. I'll try it with your instructions next time I download a new index.

[eng3]

I agree, looking at the script, that's about it but there is a little more too it.
Like having the right permissions.
I know I've tried it and sometimes it works, sometimes it doesnt.

[caddish]

I'm having trouble using a pre-downloaded photon_data, it says the folder is empty but it's definitely not

Here's the docker run

docker run
  -d
  --name='photon'
  --net='br0'
  --ip='192.168.1.34'
  --pids-limit 2048
  -e TZ="America/New_York"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Unraid"
  -e HOST_CONTAINERNAME="photon"
  -e 'UPDATE_STRATEGY'='PARALLEL'
  -e 'UPDATE_INTERVAL'='30d'
  -e 'LOG_LEVEL'='INFO'
  -e 'PUID'='99'
  -e 'PGID'='100'
  -e 'UMASK'='022'
  -e 'BASE_URL'='https://download1.graphhopper.com/public/'
  -e 'INITIAL_DOWNLOAD'='false'
  -l net.unraid.docker.managed=dockerman
  -v '/mnt/ssd/appdata/Photon/':'/photon/data':'rw' 'ghcr.io/rtuszik/photon-docker:latest'
9fac8ad01c74f1527156c3fcc5c5d76a51ac0b0f051f235d8eaa8b483718e6de

And the docker log

WARNING: Using incubator modules: jdk.incubator.vector
2025-10-20 20:45:52,613 [main] INFO  de.komoot.photon.App - Start up database cluster, this might take some time.
Exception in thread "main" 2025-10-20 20:45:52,615 [main] ERROR de.komoot.photon.Server - Data directory '/photon/data/photon_data' seems to be empty. Are you using an index for OpenSearch?
java.io.IOException: OpenSearch database not found.
        at de.komoot.photon.Server.start(Server.java:57)
        at de.komoot.photon.App.runPhoton(App.java:74)
        at de.komoot.photon.App.main(App.java:38)
Updating photon group GID from 9011 to 100
Updating ownership of files from GID 9011 to 100
Updating photon user UID from 9011 to 99
Updating ownership of files from UID 9011 to 99
2025-10-20 20:45:52,029 - root - INFO - Photon Manager starting...
2025-10-20 20:45:52,029 - root - INFO - Running initial setup...
2025-10-20 20:45:52,233 - root - INFO - No APPRISE_URLS set, skipping notification.
2025-10-20 20:45:52,233 - root - INFO - === CONFIG VARIABLES ===
2025-10-20 20:45:52,233 - root - INFO - UPDATE_STRATEGY: PARALLEL
2025-10-20 20:45:52,233 - root - INFO - UPDATE_INTERVAL: 30d
2025-10-20 20:45:52,233 - root - INFO - REGION: None
2025-10-20 20:45:52,233 - root - INFO - FORCE_UPDATE: False
2025-10-20 20:45:52,233 - root - INFO - PHOTON_PARAMS: None
2025-10-20 20:45:52,234 - root - INFO - JAVA_PARAMS: None
2025-10-20 20:45:52,234 - root - INFO - LOG_LEVEL: INFO
2025-10-20 20:45:52,234 - root - INFO - BASE_URL: https://download1.graphhopper.com/public
2025-10-20 20:45:52,234 - root - INFO - SKIP_MD5_CHECK: False
2025-10-20 20:45:52,234 - root - INFO - INITIAL_DOWNLOAD: False
2025-10-20 20:45:52,234 - root - INFO - === END CONFIG VARIABLES ===
2025-10-20 20:45:52,234 - root - INFO - Validating environment variables...
2025-10-20 20:45:52,234 - root - INFO - Environment variables are valid.
2025-10-20 20:45:52,234 - root - WARNING - Initial download is disabled but no existing Photon index was found. 
2025-10-20 20:45:52,253 - root - INFO - Starting Photon (attempt 1/3)...
2025-10-20 20:45:52,255 - root - INFO - Photon started with PID: 1843
2025-10-20 20:45:52,255 - root - INFO - Waiting for Photon to become ready...
2025-10-20 20:45:52,257 - root - ERROR - Photon health check failed after 1 attempts
2025-10-20 20:55:34,991 - root - DEBUG - Health check attempt 1 failed: HTTPConnectionPool(host='localhost', port=2322): Max retries exceeded with url: /status (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x149f9a594f80>: Failed to establish a new connection: [Errno 111] Connection refused'))

[maarten1055]

@caddish Did you solve this issue? I have the exact same problem.

[caddish]

Not really, eventually the full download completed without timing out.

I couldn't use the zip I predownloaded

I think https://r2.koalasec.org/public was timing out but https://download1.graphhopper.com/public/ worked

[Yermo]

For anyone who runs into this, I ran into the same problems. Attempting the download within the docker image yielded repeat corrupted downloads. I also tried manually downloading the database.

It took a while to realize that the correct database is the OpenSearch one, namely:

https://download1.graphhopper.com/public/photon-db-planet-0.7OS-latest.tar.bz2

It has a node_1 directory under photon_data as opposed to an elasticsearch directory.

Also the owner and group need to match what's used inside the docker image. In my case 9011:9011.