Restart clients without stopping containers

This guide will explain how to restart any CL/EL client without stopping the container. For example, you would want to stop heimdall, perform some operations in the container and then restart the client.

info

This guide assumes you have a running devnet, if that's not the case, you can head to the Getting Started section.

First, get a shell inside the first heimdall container.

kurtosis service shell pos l2-cl-1-heimdall-v2-bor-validator

Second, find the PID of the script called container-proc-manager.sh.

$ ps aux
PID   USER     TIME  COMMAND
    1 root      0:00 /sbin/docker-init -- sh -c cp /opt/data/genesis/genesis.js
    7 root      0:00 {container-proc-} /bin/sh /usr/local/share/container-proc-
   12 root      0:18 heimdalld start --all --bridge --rest-server --home /etc/h
   26 root      0:00 sh
   32 root      0:00 ps aux

In this example, the PID of the script process is 7.

Third, stop the heimdall process.

kill -s TRAP 7

Double-check that the process was effectively stopped.

$ ps aux
PID   USER     TIME  COMMAND
    1 root      0:00 /sbin/docker-init -- sh -c cp /opt/data/genesis/genesis.js
    7 root      0:00 {container-proc-} /bin/sh /usr/local/share/container-proc-
   26 root      0:00 sh
   33 root      0:00 tail -f /dev/null
   34 root      0:00 ps aux

Notice that the heimdall process with PID 12 is now gone and the container is still running. Indeed, the container proc manager script gratefully terminates the child process, heimdall, by sending a SIGTERM signal and then starts a dummy process to keep the container running.

We can now perform some operations in the container. For instance, we can clean up the database by removing stored data.

rm -r /etc/heimdall/data/*.db /etc/heimdall/data/cs.wal /etc/heimdall/data/snapshots

Then we can restart the heimdall process.

heimdalld start --all --bridge --rest-server --home /etc/heimdall