fix: propagate FAILED-state errors from inner status-check catch

TheLastCicada · TheLastCicada · commit db63b08163f9 · 2026-05-08T14:29:47.000-07:00
The inner catch block inside waitForV{1,2}OrganizationReady's status
poll only re-threw errors whose message matched 'appears stuck'. The
explicit FAILED-fast-fail throw ("Organization creation failed with
state FAILED after Xs") and the V2 'creation FAILED' branch were
silently swallowed by the same catch, never reaching the outer org-
list catch that checks for these phrases.

Bug was latent in V1 (creation-status endpoint was already correct)
and activated for V2 by this PR's endpoint switch from /status to
/creation-status. Extend the inner catch to re-throw any terminal
phrase the outer catch would have propagated, so FAILED creation no
longer waits the full 15-min maxWaitTime to surface as a generic
timeout.
diff --git a/tests/v2/live-api/helpers/live-api-helpers.js b/tests/v2/live-api/helpers/live-api-helpers.js
@@ -1307,8 +1307,17 @@ export const waitForV2OrganizationReady = async (request, orgName = null, maxWai
           }
         }
       } catch (statusError) {
-        // Re-throw stuck state errors
-        if (statusError.message?.includes('appears stuck')) {
+        // Re-throw terminal errors we explicitly threw above (FAILED
+        // state, stuck-state) so the outer org-list catch can decide
+        // what to do with them. HTTP-level errors from the status
+        // endpoint (e.g. 404 on older builds, transient 5xx) fall
+        // through and are silently logged so the loop can keep
+        // polling the org list.
+        if (
+          statusError.message?.includes('Organization creation failed') ||
+          statusError.message?.includes('appears stuck') ||
+          statusError.message?.includes('creation FAILED')
+        ) {
           throw statusError;
         }
         // Status endpoint might not exist or may fail - that's okay
@@ -1606,8 +1615,17 @@ export const waitForV1OrganizationReady = async (request, orgName = null, maxWai
           }
         }
       } catch (statusError) {
-        // Re-throw stuck state errors
-        if (statusError.message?.includes('appears stuck')) {
+        // Re-throw terminal errors we explicitly threw above (FAILED
+        // state, stuck-state) so the outer org-list catch can decide
+        // what to do with them. HTTP-level errors from the status
+        // endpoint (e.g. 404 on older builds, transient 5xx) fall
+        // through and are silently logged so the loop can keep
+        // polling the org list.
+        if (
+          statusError.message?.includes('Organization creation failed') ||
+          statusError.message?.includes('appears stuck') ||
+          statusError.message?.includes('creation FAILED')
+        ) {
           throw statusError;
         }
         // Status endpoint might not exist or may fail - that's okay
