Podman sandboxes cannot reach host.openshell.internal
Ninguém assumiu esta issue ainda.
Avaliação
- Dificuldade
- 4/5
- Tempo estimado
- 3-5 dias
- Facilidade para iniciantes
- 45/100
- Tipo de issue
- Bug
- Clareza
- Claramente especificada
- Status de atividade
- Ativa
- Stack de tecnologia
- docker, linux, rust
- Domínio
- backend, devops, networking
Direção de pesquisa
The issue is in the Podman driver's network configuration for sandboxes. Look at the supervisor's DNS setup and how the host gateway address is passed in the runtime descriptor. Start by examining the Podman driver code and the networking isolation logic. The acceptance criteria include writing an automated Podman end-to-end test, so check existing test suites for networking. 'Done' means host.openshell.internal resolves and connects in a Podman sandbox after restart, with all mediation intact.
Escrita pelo modelo de indexação a partir do texto da issue.
Descrição
User Story
As an OpenShell user running sandboxes with the Podman driver, I want workloads to access services on the host through host.openshell.internal so that host access behaves consistently across supported local
compute drivers.
Problem Statement
Podman-backed sandboxes cannot reliably resolve and connect to host.openshell.internal.
OpenShell workloads use an isolated network namespace and rely on the supervisor for mediated network access. The Podman driver does not currently provide the workload with the expected supervisor DNS
configuration or pass the trusted host gateway address into the supervisor runtime descriptor. As a result, the host alias is unavailable even when the sandbox policy permits the connection.
Impact / Why This Matters
This prevents Podman users from accessing host-local development services, test fixtures, and other explicitly permitted endpoints through OpenShell’s standard host alias.
The current workaround is to use driver-specific host addresses or change the workload’s networking configuration. That workaround is insufficient because it is platform-dependent, bypasses the portable
host.openshell.internal contract, and may weaken the intended mediated-networking boundary.
Acceptance Criteria
- A Podman sandbox can resolve host.openshell.internal.
- A policy-authorized TCP connection through that alias reaches a service listening on the host.
- The connection continues to work after the sandbox is stopped and restarted.
- Workloads continue to use isolated networking and reach the host only through supervisor mediation.
- The configured host gateway address is validated before use.
- Driver-owned resolver resources are cleaned up on failed creation and sandbox deletion.
- An automated Podman end-to-end test covers initial creation and restart.
- Relevant architecture, networking, and gateway configuration documentation is updated.
Reproduction Steps
-
Start OpenShell with the Podman compute driver.
-
Start an HTTP server bound to a host-reachable loopback address and note its port.
-
Create a sandbox whose policy permits TCP access to host.openshell.internal:.
-
From the sandbox, attempt to connect to that address:
exec 3<>/dev/tcp/host.openshell.internal/
printf 'GET / HTTP/1.0\r\n\r\n' >&3
cat <&3 -
Observe that DNS resolution or the connection fails.
The same behavior should also be checked after stopping and restarting the sandbox
Environment
- Compute driver: Podman
- Host operating system: Linux
- Workload networking: isolated (network=none)
- Connection target: host.openshell.internal
Logs
- Linguagem predominante
- Rust
- Estrelas
- 8.7k
- Forks
- 1.3k
- Merge médio
- 2d 6h
- PRs com merge (30d)
- 297
Guia de contribuição
Primeiros passos
- Leia a issue inteira e depois o guia de contribuição do projeto.
- Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
- Faça um fork do repositório e trabalhe em uma branch.
- Abra um pull request que referencie o número da issue.
Mais de NVIDIA/OpenShell
-
area:docs
Dificuldade 1/5 Menos de uma hora Facilidade para iniciantes 88/100
-
state:triage-needed
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 82/100
-
area:cli state:validated
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 72/100
-
state:triage-needed
Dificuldade 1/5 Menos de uma hora Facilidade para iniciantes 90/100
-
area:build spike state:review-ready state:stale
Dificuldade 2/5 Meio dia Facilidade para iniciantes 68/100
Todas as issues de NVIDIA/OpenShell
Issues semelhantes
-
bug github_actions
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 75/100
registrystack/registry-stack#1393 ·
-
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 75/100
longbridge/gpui-kit#3223 ·
-
bug engine
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 65/100
rocky-data/rocky#2181 ·
-
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 70/100
oasisprotocol/oasis-sdk#2523 ·
-
[indexer] [QA] Add a focused test for the new NonRetryableError / assertSocketAlive() behavior. Abertabot:ai-assisted component:indexer QA-roadmap status:untriaged
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 75/100
midnightntwrk/midnight-indexer#1557 ·