Server disk notifications

Disk (getting) full

You can use the documentation below to act on the notifications "Disk getting full", "Disk almost full" and "Disk is full".

Log in to the environment of the disk notification. Navigate to your domain folder /home/<username>/domains/<domainname>

Run the command ncdu, which will analyze the disk usage. After the scan is done, you will see output that looks similar to:

--- /home/hipex/domains/hipex.io -------------------------------------------------------------------------------------------------------------------------------------------------------------
  58.8 GiB [##########] /application
  17.0 GiB [##        ] /application-backup
   1.8 GiB [          ] /var
   4.0 KiB [          ]  redis.json
@   0.0   B [          ]  public_html

As you can see in this example, we have a lot of data in the application-backup folder. Navigate to the application-backup folder using the arrow key and press enter on application-backup folder.

--- /home/hipex/domains/hipex.io/application-backup -------------------------------------------------------------------------------------------------------------------------------------------------------
                       /..
 11.8 GiB [##########] /backup

We are going to delete this folder, press on the d key to promp the delete screen. Navigate to yes and press enter to delete the folder

┌───Confirm delete─────────────────────────────────────────┐
│ Are you sure you want to delete "backup"                 │
│                 and all of its contents?                 │
│                                                          │
│              yes   no     don't ask me again             │
└──────────────────────────────────────────────────────────┘

Disk inodes (getting) full

What are inodes?

The Linux filesystem contains objects, each object is represented by an inode. A object is file under the Linux system, for example 1 session file is 1 inode.

When you receive this notification you need to take the following steps. Check the Inodes used: df -i

Output:

[/home/hipex/domains/hipex.io]$ df -i
Filesystem      Inodes  IUsed   IFree IUse% Mounted on
devtmpfs        991422    347  991075    1% /dev
tmpfs           997404      1  997403    1% /dev/shm
tmpfs           997404    650  996754    1% /run
tmpfs           997404     16  997388    1% /sys/fs/cgroup
/dev/sda1      4780464 734745 4045719   95% /

As you can see, the Inodes on SDA are currently for 95% in use, which is quite a lot.

To print out the inode usage for the current user, you can use the following command:

{ find ~/ -xdev -printf '%h\n' | sort | uniq -c | sort -k 1 -n; } 2>/dev/null

This allows you to idenfity candidates that you can clean up to free up inode usage.

Too many session files

In many cases, a high inodes usage is caused by a huge amount of session files being stored on the server.

To solve this problem, we can clean the sessions that are older than 14 days:

find /var/domains -mtime +14 -regextype sed -regex ".*/session/sess_[a-z0-9]\{26\}" -delete && find /var/domains -mtime +14 -regextype sed -regex ".*/sessions/sess_[a-z0-9]\{26\}" -delete

After this is done, verify if the Inodes usage is below 95%, if not you can use the same command and change the value from 14 days to 7. To prevent this from happening in the future, we advise to use Redis for handling sessions..