docs: retire the source.coop outage runbook for the mirror runbook - #98
Open
cboettig wants to merge 1 commit into
Open
docs: retire the source.coop outage runbook for the mirror runbook#98cboettig wants to merge 1 commit into
cboettig wants to merge 1 commit into
Conversation
Deletes the three outage files added together in 99e49e6 (2026-07-04) as a snapshot of one headless session: headless/TESTING-DURING-CEPH-OUTAGE.md headless/TESTING-DURING-CEPH-OUTAGE.bosl.json headless/outage-systemprompt-addendum.md They describe the source.coop route: partial coverage, per-collection href remapping, and the NRP dev server for its rewrite. mcp-data-server#266 has since landed a complete drop-in mirror (minio.carlboettiger.info, identical bucket paths) plus a mirror-configured head, which needs none of that. The addendum is the reason to delete rather than update: it steers models away from browse_stac_catalog / get_stac_details / get_collection. That was correct on source.coop, which has no root catalog — but on a mirror head those tools work (verified 2026-07-30: 219 collections from duckdb-mcp.carlboettiger.info), so applying it there blinds the agent for no reason. Replaced by a 6-line pointer in headless/README.md to the single runbook, mcp-data-server docs/guide/mirror-failover.md (see mcp-data-server#347). Nothing outside these files referenced them. Merging is the call of whoever owns the headless harness — the deletion is recoverable from history either way.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Net −139 lines. Deletes three files and replaces them with a 6-line pointer.
The three files landed together in
99e49e6(2026-07-04, "outage-testing harness (#62)") as a snapshot of one headless session during that outage:headless/TESTING-DURING-CEPH-OUTAGE.mdheadless/TESTING-DURING-CEPH-OUTAGE.bosl.jsonheadless/outage-systemprompt-addendum.mdThey document the source.coop route — partial coverage, per-collection href remapping, and the NRP dev server for its rewrite (#261). Since then mcp-data-server#266 landed a complete drop-in mirror (
minio.carlboettiger.info, identical bucket names and paths, self-consistent hrefs) plus a mirror-configured head atduckdb-mcp.carlboettiger.info, which needs none of that machinery.Why delete rather than update
The addendum instructs models: "Do NOT call
browse_stac_catalog,get_stac_details, orget_collection." That was correct on source.coop, which has no root catalog. On a mirror-configured head those tools work — verified 2026-07-30 againstduckdb-mcp.carlboettiger.info,browse_stac_catalog→ 219 collections. Applying the addendum there blinds the agent for no reason, so the file is not stale-but-fixable; its premise is gone.Ceph was down again on 2026-07-30, and this was the runbook found first — which is how the lossier route got followed before the better one turned up.
Replacement
One runbook, in mcp-data-server:
docs/guide/mirror-failover.md(tightened in mcp-data-server#347, including a "For agents querying through MCP during an outage" section that covers tool choice, so no system-prompt addendum is needed).headless/README.mdgains a pointer to it.Nothing outside these three files referenced them (they only cross-referenced each other).
Note on ownership
This touches the headless harness, which isn't mine — merging is your call, and the content stays recoverable from history regardless. Flagging one thing in case it matters: if any in-flight run is currently using
TESTING-DURING-CEPH-OUTAGE.bosl.json, keep this open until that finishes. Authored from a worktree, so nothing in the shared checkout was touched.