Repository navigation
On-Premise Integration with Wazuh (v7.5.0): 503 Marketplace error & running without demo data #374
|
Hi everyone, Thanks in advance for your help! |
Replies: 1 comment
|
Hi Damian — thank you for the report, and apologies it took this long to answer. Both questions had real answers and the first one was a genuine bug on our side, now fixed. 1. The 503 was ours, and it affected every containerised deploymentYou were not doing anything wrong. The API resolves the marketplace catalogue from There was a second problem stacked behind it: install hashed the item's file under Reproduced against the published image, and re-run against the one on
The fix (#949) packages the index inside the API's build context and records each item's SHA-256 at index-build time, so install no longer needs the content trees at runtime. Worth saying plainly: nothing in CI compared what the image contains against what the code reads, and the test suite runs from a source checkout where every path resolves. That is why a 503 on a headline surface survived every release. There is now a gate for exactly that, and it fails if the packaged copy goes missing again. 2. Yes — and that is already the defaultA stock This is worth flagging because on v7.5.0, the version you were running, it was not true: the published console image had demo mode baked in at build time, so a self-hoster's dashboard announced demo resets over their own data. That was fixed in v10 — the demo build now lives under its own For an on-premise deployment I would also point you at Wazuh specificallyThe connector is first-party and does not need the marketplace at all — you can add it from Settings → Connectors in the console. Setup guide: Wazuh connector. It polls the Wazuh API, maps rule levels onto the five-tier severity ladder, and its events promote to alerts and go through auto-triage like any other source. If anything else surfaces once you are on the current image, please open an issue — this one sat far longer than it should have. |
Hi Damian — thank you for the report, and apologies it took this long to answer. Both questions had real answers and the first one was a genuine bug on our side, now fixed.
1. The 503 was ours, and it affected every containerised deployment
You were not doing anything wrong. The API resolves the marketplace catalogue from
marketplace/index.json, and it looked in four locations — all four outside the API's Docker build context. The image is built fromservices/api, soCOPY . .never saw the repository-rootmarketplace/directory and the index simply was not in the image. Browsing the marketplace page still worked, because the console serves its own static copy; only the API had nothing, wh…