Seafile-data has INSANE crazy high usage of data

(picture domain not available anymore - moderator)

As you can see in this picture, my seafile-data is using 2.7 TERABYTES of data.

Whereas, in the Seafile interface it says I am not even hitting 800 GIGABYTES!

What is causing this INSANE difference that is now not allowing me to upload anything to my server?

I ran in to this issue back then and I always see new users run into this same learning. This topic should be mentioned more obviously in the setup guide. So that people consider setting up GC right from the beginning.
I have to look at the manual, how and if it is mentioned in the installation part

1 Like

It would be nice if we could have a report on how much of data is in trash by user and library.


I don’t think it is a matter of garbage collection.
I have the same problem (1,8 To used seen from the admin panel, 3,2 To seen from the system…).
We have scripts running automatically that do garbage collection and it doesn’t change anything.

I think I might have read in this forum that the file history is not considered when used spaced is measured.

For me these issues are always a garbage collector problem until people say they have just run it. But never seen people come back with this after running GC…

There must be a begining for anything :slight_smile:

I have a crontab job running automatically -r and then every thursday using Seafile user.
The script’s output is sent to a log file.
If I take a look at the latest logs I can see this kind of thinks :
[08/08/19 05:50:56] gc-core.c(584): GC finished for repo 1393e514. 10560 blocks total, about 10536 reachable blocks, 34 blocks are removed.
If you thinks it is related to a GC misuse, what do I miss ?
Is there a giant trash folder hidden somewhere I didn’t find yet ?

History is not accounted for in sysadmin panel. How much history you have and which old items are removed by the GC depends on your versioning settings.

exactly, this would have been my next advice. Because if your versioning settings are like 100 days or “keep all” your GC will never throw away that stuff. I have no Repo with “keep all/forever”, its to risky for me.

And thanks for your answers.
As I previously wrote :

We are hosting a multi-tenant version of Seafile Pro, which means that our customers are free to set the history length as they want. Then if the used space doesn’t include their file history, it also means that we cannot bill them for that and that we are the ones paying the bill for their use… :sleepy:
So GC cannot do anything for that matter…
@daniel.pan would it be possible to measure the real used space ?

There are settings to set default keep_days for all the libraries and let users don’t be able to change history setting. Is it possible for you to turn on the options.

It is not possible to calculate the space of all the histories because it is time consuming.

But could he as a service provider run these time-consuming tasks somehow by his own and accommodate this data?
Doesn’t ever library have a own folder? The amount of space it consumes is the correct size with history, isn’t ? What is the Problem going of the file system and the then checking the library sizes? Can’t you map the quirky library I’d to the user?

This works if you use file system as storage backend to count the size of the folder for each library. But it is slow too.