Canonical answer to “where do I build/test FaceKom natively?” — the fk-dev Tailscale VM. The old on-prem ssh Facekom box is decommissioned. Verified 2026-07-01.
ssh Facekomis DEAD — do not resurrect itThe old host (
Host Facekom→HostName localhost,User lederera,ProxyJump FKJumpBox→root@lederera-447-fk-hardver) is offline since ~2026-06-27 and decommissioned. Tailscale reportslederera-447-fk-hardveras offline. Per the user: “never use the ssh Facekom host again, we need to use the tailscale fk vm”. Any doc still saying “build onssh Facekom” meansfk-dev.
For Agents
- Host:
fk-dev— GCP VM on tailnettaild4189d.ts.net, tailnet IP100.91.108.61, MagicDNSfk-dev.taild4189d.ts.net.- Connect:
command ssh ops@fk-dev.taild4189d.ts.net(userops, Tailscale SSH — no keypair).sshgotcha: the user’s interactive shell aliasesssh/scpto a_kaku_wrapped_sshfunction that is NOT loaded in a non-interactive shell → plainsshfails with_kaku_wrapped_ssh: command not found. Always usecommand ssh/command scpfor the raw binary.- Use a login shell for remote docker work:
command ssh ops@fk-dev.taild4189d.ts.net bash -l -s <<'EOF' … EOF.- Still true: build native x86_64 on the remote host, NOT emulated on the Mac — qemu/podman emulation gives SIGSEGV (exit 139), overlay-FS I/O errors (exit 125) and repo-metalink 503s. Only the host changed.
- Operational runbook (deploy a branch by bind-mount, no image rebuild): fk-dev-deploy-smoke-runbook (general recipe + SSH policy + OpenHours setup) / fk-dev-partner-branch-deploy-runbook (partner swap — uid 1001-vs-1000 ownership traps, the read-only per-service
config/local.json, per-partner DB, operator users, css asset build) / fk-dev-nusz-deploy-and-8959-verification (NÚSZ/8959-specific). Architecture/provisioning: tailscale-gcp-dev-box-migration.
Which SSH users work — tested 2026-08-12
opsandrootboth work. The documentedcommand ssh ops@fk-dev.taild4189d.ts.netabove is correct; the shortops@fk-devandroot@fk-devforms work too.
User Result opsOK rootOK levander,ubuntu,admin,facekom,vuer,dev,techteamerrefused — tailnet policy does not permit you to SSH as user "<u>"
rootre-confirmed working 2026-08-18; the extra refused users (vuer,dev,techteamer) were tested then too — see Access. The per-service names (oss-fk-dev,css-fk-dev, …) are Tailscale sidecar containers, not SSH hosts — port 22 is refused there.Connect by name, not by IP.
fk-devandfk-dev.taild4189d.ts.netare both inknown_hosts;100.91.108.61is not, and the IP form fails withHost key verification failedeven though the IP is correct.
Hosts
| Host | Status | Reach |
|---|---|---|
fk-dev (100.91.108.61) | CURRENT — build/test/deploy here | command ssh ops@fk-dev.taild4189d.ts.net |
lederera-447-fk-hardver (ssh Facekom / FKJumpBox) | DECOMMISSIONED — offline since ~2026-06-27 | — |
Tailnet peers (per-service sidecars on fk-dev)
oss-fk-dev (100.91.55.42), css-fk-dev, portal-fk-dev, esign-oss-fk-dev, esign-css-fk-dev, esign-api-fk-dev, css-sdk-demo-fk-dev — each a per-service Tailscale sidecar with its own MagicDNS name (e.g. operator UI at https://oss-fk-dev.taild4189d.ts.net). They are not separate build hosts; the docker stack runs on fk-dev itself.
What this replaces
Everything previously done “on ssh Facekom” now happens on fk-dev:
- Native image builds — UBI10 migration work (FKITDEV-8252),
docker compose buildof vuer_docker / vuer_build / vuer-release images. - Release packaging — vuer_build
./build.sh -b <client>-<ver>+./sign-partner.sh(see release-automation-design). - Runtime/branch testing — bind-mounted
/workspace/<repo>checkouts, in-containeryarn install && yarn build,supervisorctl restart all.
Historical notes
Investigation notes written before 2026-07-01 (e.g. dev-box-esign-container-startup-failures-2026-06-01, instacash-external-api-esign-headless-test-2026-06-01, FKITDEV-8787, instacash-esign-dev-box-deploy) describe sessions run on the old box. Their findings (container/image gotchas, deploy recipes,
command sshtrick) still hold — only substitute the host withfk-dev.