Version: v5.7.0

opschain admin server site

Included from Basic tier

The per-site operations verbs for multi-site estates: durably drain EVERY server node of one site so the whole site can be patched, rolled or taken down together (quiesce), see which sites are held (list), and bring a site back with an explicit operator act (resume). Sites are addressed by the site identity the servers carry in their site config key (or OPSCHAIN_SITE) — the same vocabulary as the reserved location runner-group attribute (e.g. dc1).

A quiesced site's nodes each drain exactly as a per-node quiesce drains one node (opschain docs cli/admin-server-node): readiness off, job offers stopped, scheduler participation handed back, in-flight run drives detached for the other site's peers to adopt. The site's servers are then safe to kill and roll together. The site's RUNNERS are not stranded: they fail over to the next server in their ordered list (opschain docs cli/runner), park on the surviving site, and still obtain their own site's work there — another site's servers claim a quiesced site's steps on its behalf, while placement rules keep deciding which runners may execute them (opschain docs dispatch).

The posture is durable: a node of the site restarted under it boots HELD (healthy and inspectable, taking no work) until resume. There is deliberately no timer or automatic rejoin. And the hold is never advisory: a node that cannot READ the site posture stops offering work rather than guessing.

Subcommand	Purpose
`list`	Show the quiesced sites, and the answering node's own site and held state — `opschain docs cli/admin-server-site-list`.
`quiesce <site>`	Durably drain every node of one site — `opschain docs cli/admin-server-site-quiesce`.
`resume <site>`	The explicit human resume a held site waits for — `opschain docs cli/admin-server-site-resume`.

See also​

See also