Provider OpenToFu creating zombie git processes

[alexmeise]

What happened?

During normal operations provider-opentofu creates git processes that are staying in a defunct state indefinitely (until the Pod is restarted).

As a result, it creates a zombie processes that should not exist.

A few hours with a small amount of resources accumulated ~350 zombie processes. With more resources during a weekend it can go well over 10 0000.

It can be seen either by running ps from a Node or a Pod:

• Pod: kubectl exec -it pod/PROVIDER-OPENTOFU-POD -- ps

• Node: ps -afx

Example output (reduced):

> ps -afx
17303 ?        Ssl    0:06  \_ crossplane-opentofu-provider
17755 ?        Zs     0:00      \_ [git] <defunct>
17967 ?        Zs     0:00      \_ [git] <defunct>
18115 ?        Zs     0:00      \_ [git] <defunct>
21050 ?        Zs     0:00      \_ [git] <defunct>
21117 ?        Zs     0:00      \_ [git] <defunct>
21209 ?        Zs     0:00      \_ [git] <defunct>
23216 ?        Zs     0:00      \_ [git] <defunct>
23414 ?        Zs     0:00      \_ [git] <defunct>
23540 ?        Zs     0:00      \_ [git] <defunct>
25333 ?        Zs     0:00      \_ [git] <defunct>
25448 ?        Zs     0:00      \_ [git] <defunct>
25535 ?        Zs     0:00      \_ [git] <defunct>
27277 ?        Zs     0:00      \_ [git] <defunct>
27619 ?        Zs     0:00      \_ [git] <defunct>

How can we reproduce it?

Create any instance of a Workspace with a remote source and check the dead processes piling up. Excerpt from an actual Workspace used below:

module: git::https://<--REDACTED-->
source: Remote

What environment did it happen in?

Crossplane v1.18.1
Provider OpenToFu v0.2.5
Rancher Desktop: v1.19.3
K8S version: v1.32.6 (issue was present on other versions also)

[fbozic]

Are there any updates about this issue? We are facing the same issue.