I just created a new box that I put Virtualmin on. Fresh minimal install of Debian 11. Debian had PHP 7.4 on it so I added PHP 8.0, 8.1, and 8.2 to it.
I am curious about the accuracy of the “Disk usage” in the notification bar. The notification indicates this:
Mounted As ---- Free --------------------- Total
/ ---------------------- 95% (8.55 TiB) ----- 9.02 TiB
/boot -------------- 74% (346.5 MiB) – 470.35 MiB
The box has no websites other then the default box name and a couple megabytes there to serve up a default page.
Some math calculations tell me, based on the indicated “Disk usage” that the space consumed on the box is some 470 gigabytes. That seems to be way off for a new box.
I did this from / on the box:
du -h
and got a result of 8.6 G which actually makes sense.
Out of curiosity, I then checked my original Debian 10 box with Virtualmin on it. The “Disk usage” notification area on that one indicates this:
Mounted As ---- Free ---------------------- Total
/ ---------------------- 89% (194.52 GiB) - 220.31 GiB
/boot -------------- 36% (83.54 MiB) – 233.32 MiB
Doing this from / on that box:
du -h
yields a result of 18 G which actually makes sense because it has a number of websites on it.
It was:
du -h
I did on both boxes to compare against what the Virtualmin/Webmin notification panel indicated. And yes, the second box I checked disk usage on had 89% free.
It didn’t seem feasible that a freshly loaded box was using almost 500 gigabytes of disk space.
OK. Definitely something off with disk usage reporting. I wrote a script to move emails. Went into the interface to check and had a cow. This user is an old friend and I don’t think he’s ever deleted an email. But, I knew going in how big his account was. When I went in and looked I saw:
That’s three times the size it should be? I had created the accounts a couple weeks ago and was using it to test the transfer. Had rsync really created dupes? My account looked normal.
Yeah, it appears to be double (or triple/quadruple/whatever) tmpfs filesystems, or something. I see a minor discrepancy, as well. I’ll ask @Ilia to take a look.