Che deployment should not fail because pulling images took too long

[sleshchenko]

Is your task related to a problem? Please describe.

I'm not sure but I suppose that it makes sense to increase pod wait timeout in chectl.
I'm raising this question because when I deploy Che on CRC I got:

~/projects > chectl server:start --platform=crc                                                                                                                50s 15:22:27
› Current Kubernetes context: 'default/api-crc-testing:6443/kube:admin'
....
  ❯ ✅  Post installation checklist
    ✔ PostgreSQL pod bootstrap
      ✔ scheduling...done.
      ✔ downloading images...done.
      ✔ starting...done.
    ❯ Keycloak pod bootstrap
      ✔ scheduling...done.
      ✔ downloading images...done.
      ✖ starting
        → ERR_TIMEOUT: Timeout set to pod ready timeout 130000
      Devfile registry pod bootstrap
      Plugin registry pod bootstrap
      Eclipse Che pod bootstrap
      Retrieving Eclipse Che server URL
      Eclipse Che status check
    Retrieving Keycloak admin credentials
    Retrieving Che self-signed CA certificate
 ›   Error: Error: ERR_TIMEOUT: Timeout set to pod ready timeout 130000
 ›   Installation failed, check logs in '/tmp/chectl-logs/1600863851534'

which is kind of OK because keycloak is started in 120107ms ( which is less than the default 130000 timeout but if we add image pulling here - it exceeded timeout)

12:25:24,089 INFO  [org.jboss.modules] (CLI command executor) JBoss Modules version 1.9.0.Final
12:25:24,927 INFO  [org.jboss.msc] (CLI command executor) JBoss MSC version 1.4.5.Final
12:25:25,031 INFO  [org.jboss.threads] (CLI command executor) JBoss Threads version 2.3.3.Final
...
12:28:30,407 INFO  [org.jboss.as.server] (Controller Boot Thread) WFLYSRV0212: Resuming server
12:28:30,422 INFO  [org.jboss.as] (Controller Boot Thread) WFLYSRV0060: Http management interface listening on http://127.0.0.1:9990/management
12:28:30,425 INFO  [org.jboss.as] (Controller Boot Thread) WFLYSRV0051: Admin console listening on http://127.0.0.1:9990
12:28:30,426 INFO  [org.jboss.as] (Controller Boot Thread) WFLYSRV0025: Keycloak 6.0.1 (WildFly Core 8.0.0.Final) started in 120107ms - Started 617 of 880 services (563 services are lazy, passive or on-demand)

Describe the solution you'd like

The simplest way to help me is just increasing the default value for timeout. But below see maybe even better alternatives:

Describe alternatives you've considered

1. We reset timeout when pod waits get to next step, so we have different timeout for pod scheduling, image pulling and containers start up.
2. Waiting process is interactive by default which means when we exceed timeout - we ask user is they want to keep waiting or interrupt waiting.

• in my particular case I would wait because I need keycloak admin password which is printed only when Che is fully started.

Additional context

[l0rd]

Another option is not to use a timeout at all and double check for errors instead (in events, in operator, in pod status). Timeout is frustrating if something fails (we need to wait for 2 minutes when we could have been notified imediately) and is frustrating when everything went well but pulling images took too long (we tell the user that the deployment was a failure when it actually worked fine).