In [this epic](https://gitlab.com/groups/gitlab-org/-/work_items/20030), we are working on adding configurable client retries to all Gitaly clients. This is needed to release Gitaly on Kubernetes in General Availability (GA). To provide zero downtime upgrade with Gitaly on Kubernetes, we use the client retries as a mechanism to avoid returning an error to clients: the idea is to configure the client retries with enough room to _patch over_ the time it takes for a Gitaly Pod to start. The following clients have already been updated (or the work is in progress): * Rails * Workhorse * GitLab Shell We recently realized that the KAS project is also a client of Gitaly. This issue is to track the discussion about the needs (or not) to introduce configurable client retries in the [Gitaly client here](https://gitlab.com/gitlab-org/cluster-integration/gitlab-agent/-/blob/master/internal/cmd/kas/command.go?ref_type=heads#L566).