Where to run back up command?

[greyfence]

I'm trying to back up the database before doing the upgrade to v5 but the method in the docs aren't working for me.

I'm trying to do the 'docker compose exec postgis pg_dump...' method but when I run the command on the base OS I get a 'service postgis is not running' and when I try to run it in the container itself I get a 'docker doesn't exist' error.

Where am I supposed to run that command? Is there not an easier way to do a back up??

[dgraf-gh]

Hello @greyfence,

the commands in the documentation assume you are running this in the same folder where your docker-compose.yml lies on the host system. If you are running them inside the container, just drop the docker compose exec postgis part before the command. But then you would need to copy the dump out of the container somehow.

Maybe we can create a dedicated backup feature which either does bundle them automatically somewhere configurable.

[greyfence]

I would very much like a way of backing things up from within the app itself.

I did try running the full 'docker compose...' command from the same folder as the compose.yml. (I'm using portainer and it took a while to find where that was!). That's where I was getting the 'postgis is not running' error.

I'll try the shorter command in the container and see if I can use wget to upload it to my server.

Thanks.

[greyfence]

OK, I took a risk and did the upgrade without figuring out how to do the backup. It did not go well. I now have spurious locations in most days trips.

Here's an example: before the upgrade this trip showed the line almost perfectly following my route. Now it has all these random points randomly into the middle of nowhere.

Can this be fixed?

[dgraf-gh]

Hello @greyfence,

we will figure that out!

From a first look at the screen, it seems like it is not filtering out suspicious points.

A gps point in reitti can have 3 status.

• 0: normal
• 1: deleted by you
• 2: ignored by the system

1 and 2 should not be displayed.

For a first try, you can got to the day in question, then in main menu open the Workbench. It should bring you to that day then. Select one point and move it around by a little bit. Save it and let it recalculate that day. If that helps, I will find a way to do it automatically. But first we need to verify that.

Let me know how it goes.

[greyfence]

I'll try that test. I've never deleted a point so there shouldn't be any '1's at least.

For the moment I've reverted to a backed up copy of my docker VM. I didn't want to lose too many days worth of data in case I had to revert later. I exported the gpx from that day in both the new version and old version and they're identical so I guess based on that and what you're saying is that it's just a display issue rather than a messed up data issue?

For comparison, here's what the same trip looks like in the older version of reitti:

[dgraf-gh]

Yes, I do not think that any data is lost. It is probably some issue how old ignored points are transferred into the new structure. I am a bit curious why you are the first one to notice that, I can not see that on my instance.

When you export the gpx data, the flags (ignored in particular) are ignored because we want to export the raw data which came into reitti. For the calculation and display we will ignore them.

[greyfence]

I booted back into the VM with the new version and moved some points closer to where they should have been and clicked save ... is that what causes it to recalculate the day? I didn't see any other point auto-magically move into their correct position.

You can see here that the weird triangle of points is gone after I moved them but none of the other wrong points have moved:

[greyfence]

One other little tidbit: I had the new version running for two days and wanted to import the points from those two days into the older version. I exported the gpx from the new version, booted into the old version and imported that gpx.

When I view the map it is showing the odd point spikes like the new version was. I thought the gpx data would have been the 'clean' data from Owntracks and it was just the interpreting of the points that had changed from one version to the next?

[dgraf-gh]

No, that is another problem.

GPX on its own does not have an notion about accuracy or quality of the point in question, so the 4.0 importer just assumes an accuracy of 10m. Therefore the importer does assume that is a valid (in the sense of reitti) point. It is actually the point Owntracks reported into reitti.

Beginning in 5.0 we utilize an extension mechanism in the GPX file to add that information. But the 4.0 importer does not know about it.

That is the reason these spikes are back for these 2 days. I am still trying to figure out why this happened to you. At least you are the only one reporting it, and I did not notice that on my machine.

The best you can do right now is staying on 4.x and when I figured something out, move to the newest 5.x version. You can fix these two days then later.

[greyfence]

OK, thanks. Let me know if you need me to try anything.