fix docker stop after 10s

[xiagw]

No description provided.

[deitch]

I don't mind. What does it gain us?

[xiagw]

Securely exit the program and save time.

[deitch]

I guess. Make entrypoint part of the ENTRYPOINT, please. It could be possible to add other arguments that should be passed to it

[galti3r]

Work for me.

Without PR

❯ docker compose exec mysql-backup bash
87ebbd66780c:/$ ps
PID   USER     TIME  COMMAND
    1 appuser   0:00 {entrypoint} /bin/bash /entrypoint
    9 appuser   0:00 sleep 60
   10 appuser   0:00 bash
   16 appuser   0:00 ps
87ebbd66780c:/$ 
exit

~/Repositories/Kadiska/mysql-backup on master
❯ docker compose restart
[+] Restarting 3/3
 ✔ Container mysql-backup-mysql-1         Started                                              2.1s 
 ✔ Container mysql-backup-wordpress-1     Started                                              1.5s 
 ✔ Container mysql-backup-mysql-backup-1  Started                                             10.4s 

Container mysql-backup-mysql-backup-1 Started 10.4s

With PR

❯ docker compose exec mysql-backup bash
a2cd172bfea6:/$ ps
PID   USER     TIME  COMMAND
    1 appuser   0:00 /usr/bin/dumb-init -- /entrypoint
    7 appuser   0:00 {entrypoint} /bin/bash /entrypoint
   22 appuser   0:00 sleep 86400
   29 appuser   0:00 bash
   35 appuser   0:00 ps
a2cd172bfea6:/$
exit

~/mysql-backup on master !1 ?1
❯ docker compose restart
[+] Restarting 3/3
 ✔ Container mysql-backup-mysql-backup-1  Started        0.4s
 ✔ Container mysql-backup-wordpress-1     Started        1.6s
 ✔ Container mysql-backup-mysql-1         Started        1.8s

Container mysql-backup-mysql-backup-1 Started 0.4s

[deitch]

Looks fine to me. Awaiting a rebase and fixing the ENTRYPOINT/CMD

[galti3r]

Looks fine to me. Awaiting a rebase and fixing the ENTRYPOINT/CMD

About dumb-init
There is multiple approach :

• (Good solution) Manage SIGNAL traps directly inside bash program
• (Good workaround solution with more features inthebox) Use dumb-init as ENTRYPOINY then load CMD command. Actual entrypoint file can be renamed run.sh for easiest understanding. Also self traps management could be a good solution to advertise any restart during dump or restore. Actually after 10s (by default), program will be killed, even if restore is processing.
• (Crapy solution) change docker-compose container timeout, with stop_grace_period: 1s for example