| Previous | Next |
| NS_E_CACHE_ORIGIN_SERVER_TIMEOUT | NS_E_CACHE_CANNOT_BE_CACHED |
NS_E_CACHE_NOT_BROADCAST
NS_E_CACHE_NOT_BROADCAST: owning object, evidence and recovery
Where the boundary sits
This result (0xC00D13EF) marks cache policy requires broadcast content but the origin metadata identifies another stream type. The code identifies a narrow server contract, so diagnosis should preserve the object generation and the first native return.
The cache/proxy layer first applies policy, then decides whether to serve local content, revalidate it, contact an origin server, proxy it, redirect the client or reject the request., those decisions are separate from ordinary player buffering. In a this result trace, keep the cache key, origin validators and policy decision together. While diagnosing it, a cache proxy behaves both as a server to the player and as a client to the origin. The decisive question is whether the live object and values match that boundary; the base AllStat description alone does not reveal the object generation, selected plug-in or lower-level failure.
Interpret the retest
| Retest observation | Interpretation |
|---|---|
| The same call still returns this result | The rejected precondition has not changed, or the caller is still using an old object/configuration generation. |
| The operation advances and a later code appears | The boundary was cleared. After this result, diagnose the new code at its own source, parser, sink, network or client stage. |
| A new object succeeds while the retained object fails | Object lifetime or stale context is part of the incident; update lifecycle handling rather than applying a machine-wide repair. |
| Only one publishing point, playlist, cache key or plug-in fails | The evidence favors object-specific configuration or content over a server-wide outage. |
Code-specific failure anatomy
In a representative incident, the server reaches cache policy requires broadcast content but the origin metadata identifies another stream type and rejects the operation before the caller can safely assume the next stage occurred. The incident record should therefore join origin response headers, x-wms-stream-type or equivalent content metadata, publishing-point type and requested cache action with the object generation and the exact administrative or protocol request.
A useful negative control is test with a true broadcast publishing point or choose a cache action valid for on-demand content. If that change advances the same it call, the result supports this boundary. If it remains, return to the first lower-level event instead of broadening the repair.
The tempting but misleading response is renaming the URL to look like a broadcast stream. That action does not test the distinction that matters here: cannot be cached is a policy prohibition; not broadcast is a mismatch with an operation specifically requiring broadcast semantics. This distinction is also why monitoring should retain the symbolic name instead of storing only a generic COM failure.
What to record before changing anything
| Evidence | Why it matters for it |
|---|---|
| Decisive state | origin response headers, x-wms-stream-type or equivalent content metadata, publishing-point type and requested cache action. |
| Owning object | Record the server, publishing point, playlist, namespace node, plug-in or cache item that returned it, including its creation or restart time. |
| First lower-level result | Preserve the earliest Win32, socket, COM, parser or plug-in event before the HRESULT; later wrappers can map several causes to it. |
| Controlled comparison | Use a known-good object of the same type and vary only the precondition described as “cache policy requires broadcast content but the origin metadata identifies another stream type”. |
| Security-sensitive data | Log identifiers, lengths, hashes and redacted URLs where possible; do not publish passwords, authorization files or unrestricted client data. |
Smallest useful experiment
- Capture
0xC00D13EF, it, the exact API/administrative action and the first failure timestamp. - Preserve origin response headers, x-wms-stream-type or equivalent content metadata, publishing-point type and requested cache action.
- confirm that the object still belongs to the current WMServer, publishing-point or presentation generation.
- Perform one isolated experiment: test with a true broadcast publishing point or choose a cache action valid for on-demand content.
- Repeat the original the operation through the same protocol and service account; do not substitute a different client-side test.
- After it, verify the expected next state and retain any later HRESULT as a separate pipeline result.
Success means the same operation crosses this checkpoint and produces the expected next state, not merely that the symbolic code disappears.
Do not merge nearby HRESULT values
Primary distinction: cannot be cached is a policy prohibition; not broadcast is a mismatch with an operation specifically requiring broadcast semantics.
| Nearby result | Different checkpoint |
|---|---|
NS_E_CACHE_CANNOT_BE_CACHED | Compare its own symbolic boundary and the first failing call; it must not be grouped automatically with it. |
NS_E_CACHE_ORIGIN_SERVER_TIMEOUT | Relative to it, this neighboring result belongs to another state or validation branch even when the user-visible symptom is similar. |
NS_E_CACHE_NOT_MODIFIED | Use the object type and operation sequence to determine which result is authoritative. |
Changes that do not establish the cause
- renaming the URL to look like a broadcast stream.
- a broad reinstall is especially weak evidence here because it changes many unrelated components while leaving the rejected server precondition unexplained.
- Do not suppress it or replace it with a generic “media server error”; retain the symbolic code and owning operation in telemetry.
Technical references
- Windows Media Server Cache Proxy Functionality
- Creating Cache Proxy Plug-ins
- Windows Media Services HTTP Proxy Server Interaction
- Microsoft HRESULT registry
Close the incident only after it clears on a current object.
Looking for a different code? Search another status or error code.