5.0 Beta Discussion

[dgraf-gh]

If you find something unusual, something hard to understand, something buggy or just have any ideas for improvement before the go live of 5.0, add them here.

Any feedback is welcome ❤️

[Voxel07]

Awesome changes for 5.0. The multi-device support and the workbench are exactly what I was waiting for. I did not upgrade yet, so I could only read the documentation. Is there a limit on how many additional devices can be added?

[dgraf-gh]

Hello @Voxel07,

there is no limitation on the amount of devices. At one point the timeline on the left will propably need some work if you want to display them all at once on the map. Right now it uses the user switcher to also display device avatars. If there are to many we need to work out a ui for that.

[Sblop]

With the latest beta I've started running into some problems.

I've been using Reitti since the early releases. I update my location every 20 seconds while moving and every 2 minutes while stationary.

It seems like the number of points in the database has exploded.

In raw_location_points I currently have about 6.2 million rows. Around 840,000 have synthetic = false, while the remaining 5.4 million are synthetic = true.

I'd like to reprocess all points, so I'm trying to reset processed = false for every row, but updates are very slow with a table this large.

Is it safe to delete all rows where synthetic = true, or are they needed for later processing?

While investigating the logs, I also noticed repeated duplicate key errors. The following insert occasionally fails because a row with the same (user_id, timestamp) already exists:

• (2, 2026-06-30 09:23:39+02)
• (2, 2026-06-30 09:27:39+02)
• (2, 2026-06-30 09:28:09+02)

I wonder if the same time window is sometimes processed more than once, or if multiple jobs can process the same user concurrently. I also noticed the query uses:

timestamp >= ? AND timestamp <= ?

Could that allow overlapping windows if one batch starts at the exact timestamp where the previous one ended?

Would it make sense to make the insert idempotent (for example with ON CONFLICT DO NOTHING), or is the duplicate processing itself the bug that should be fixed? Right now it looks like the duplicate insert causes the entire job to fail.

[dgraf-gh]

Hi @Sblop,

there is currently a bug when uploading multiple gpx file for example at once. It does not have to do something with the gpx files More a problem that the jobs do not run sequentally anymore. I am currently in the process of fixing that.

I believe the duplication comes from that, else your point is also a good starting point. I also noticed it locally.

For the other part, to support the visit detection these synthetic points are inserted. We strive for a point every 15 seconds right now. It is buried in the configs with no easy way of adjusting this. Since you log every 20 seconds, another point will be inserted always.

I could add a flag to adjust that number, then this should get rid of most of them.

There is a way of recalculating everything but I would advise against that right now. Let me finish the next (and hopeful final) beta release.

[dgraf-gh]

Hi @Sblop, the fix landed in the develop tag. This will be the final version. You are free the switch to that.