Skip to main content

Scheduling tasks

AHD VM uses Systemd as an init system.

You can leverage systemd in order to schedule periodic tasks.

A potential use case is scaling the number of replicas down to 0 for the restapi deployment during a certain maintenance period (Sunday 00:00:00 - Sunday 01:00:00)

1. Choose a name for the task​

In this example, we'll use the name restapi-down

2. Create a timer file​

Connect to AHD VM using SSH.

Using the nano text editor, create a file named restapi-down.timer

sudo nano /etc/systemd/system/restapi-down.timer

Inside the file, paste the following content:

[Unit]
Description=

[Timer]
OnCalendar=
Persistent=true

[Install]
WantedBy=timers.target

Fill out two fields:

  • Description: short description about the task
  • OnCalendar: calendar event in which this timer is activated
warning

You must use the following format to express the calendar event: DayOfWeek Year-Month-Day Hour:Minute:Second in UTC time (-2h w.r.t Italian time)

For example: since the maintenance window starts on Sunday 00:00:00 AM (Italian time):

[Unit]
Description= Scale down restapi deployment

[Timer]
OnCalendar=Sat 22:00:00
Persistent=true

[Install]
WantedBy=timers.target

3. Create a service file​

Using the nano text editor, create a file named restapi-down.service

warning

Use the same name for the .timer and .service files

sudo nano /etc/systemd/system/restapi-down.service

Inside the file, paste the following content:

[Unit]
Description=
After=network.target

[Service]
Type=oneshot
ExecStart=

Fill out two fields:

  • Description: short description about the task
  • ExecStart: command(s) that are executed when the service starts

For example: scale the number of replicas to 0 for a restapi deployment

[Unit]
Description= Kubectl command to scale down a restapi deployment
After=network.target

[Service]
Type=oneshot
ExecStart=kubectl scale deploy data-restapi --replicas=0
info

You can find the deployment's name using:

kubectl get deployment

4. Reload systemd manager configurations​

sudo systemctl daemon-reload

5. Enable and start the timer unit​

First

sudo systemctl enable restapi-down.timer

Then

sudo systemctl start restapi-down.timer

6. Check that the timer unit is active​

systemctl status restapi-down.timer

You should see:

  • the Trigger field, showing the time of the next schedulation
  • the Triggers field, showing which service to trigger (in this case, the restapi-down.service)
● restapi-down.timer - Scale down restapi deployment
     Loaded: loaded (/etc/systemd/system/restapi-down.timer; enabled; preset: disabled)
     Active: active (waiting) since Thu 2024-08-08 10:01:03 UTC; 24h ago
    Trigger: Sat 2024-08-10 22:00:00 UTC; 1 day 11h left
   Triggers: ● restapi-down.service

7. View logs of the systemd service and its execution history​

journalctl -u restapi-down.service

The above example only shows the creation of a scheduled service for scaling restapi to zero.

To setup a maintenance window properly we must not forget about scaling the restapi deployment back to 1 by creating two additional files:

  • restapi-up.timer
  • restapi-up.service
sudo nano /etc/systemd/system/restapi-up.timer
[Unit]
Description= Rescale restapi

[Timer]
OnCalendar=Sat 23:00:00
Persistent=true

[Install]
WantedBy=timers.target
sudo nano /etc/systemd/system/restapi-up.service
[Unit]
Description= Kubectl command to scale a restapi deployment to 1 replica
After=network.target

[Service]
Type=oneshot
ExecStart=kubectl scale deploy data-restapi --replicas=1