Quantcast
Channel: Debian User Forums
Viewing all articles
Browse latest Browse all 3567

General Questions • The case of the missing storage (includes solution)

$
0
0
Recently I cloned a replication server. This server does nothing other than sit and monitor its parent, ready to swap itself into primary position if the main server goes down. Periodic rsyncs keep it up to date.

After I finished and got everything running, I noticed that df showed me 10GB more space used on the boot drive of the replication server than on its parent. This should not be; the contents should be pretty close to identical. So I fired up ncdu, which didn't display the problem; it only showed up with df. Poking around in ncdu, I couldn't find the missing space.

The answer (which I eventually stumbled on more or less accidentally) turned out to be a mount issue. Each of these servers has a RAID1 mounted for storage. At some point in the cloning process, I must have run a replication job without the RAID mounted, and 10GB of mail got copied to the bare mount point. Later, when the array was mounted, the 10GB of mail on the bare mount point became invisible to ncdu, but not to df, which still registered it as used space.

Obviously I went back to my replication job and added a check to make sure the array is mounted before writing to that location. But it was a useful lesson.

Statistics: Posted by dwasi — 2024-04-02 18:33 — Replies 0 — Views 15



Viewing all articles
Browse latest Browse all 3567

Trending Articles