Icecast Streaming Media Server

[square_eyes]

I have been told they applied the patches from Github. Although my version still shows 2.3.2-kh31.

Either way the problem persists.

[square_eyes]

I have had a special character '!' in my passwords. The hosting company say this might have caused the issue. Karl do you know if this is feasible?

Note that in my proof of concept I used the same passwords and it worked fine on the windows 2.3.2 build.

[karlH]

I don't see why a single ! would be an issue. if you had something like <! unescaped then it could cause an xml parsing issue, but that would show up in the error log.

If changing to something with all alphanumeric characters makes it work then it would narrow the issue down. This is why getting logs is very useful.

karl.

[square_eyes]

After asking many times it turns out logging also died. They killed the PID (assumed that meant icecast service. They said ices, but I didn't think I was using that) and it didn't come back up. It wasn't accepting any source connections. They blamed me for using non alpha-num char in my password.

I have been moved over to a vanilla config, and lost all my reporting. At least they kept my IP and port.

Back to the drawing board. Now I'll re-test in prod. And I should have access to my own logs now.

[square_eyes]

OK here goes. I only pasted logs for cutting /autodj encoder, falling back to fallbackfile.mp3 and failing to come up to /live. This is to reduce the post size. Let me know if you need to see me attempting to reconnect /autodj.

Keep in mind this is a new prod setup. Hopefully I haven't introduced any new problems. This time my fallback file was white noise and I could hear myself get stuck on it. Interestingly I wasn't dropped this time. Also note I have changed my passwords to alpha-num and the hosting co. hasn't increased my queue size yet.

Is this the culprit?

[2013-02-14 18:46:37] INFO admin/admin_handle_request Bad or missing password on mount modification admin request (metadata)

---

[karlH]

no, that is just a rejection of a metadata update, most likely due to the source not running yet.

The problem should now be fixed in github. The routine for overriding listeners still used a legacy API which worked with 2.3.x, however in KH, the fallback files are treated differently internally and the override was skipping them if they were cascaded, ie live ->autodj->file.

I've now avoided that routine and cascade properly and the listeners now get moved.

karl.

[square_eyes]

So are you saying you have just updated something now... and I need my hosting provider to apply that?

Can you please tell me exactly which patch/version to use?

Remember they updated the Github patch and the issue was still there.

[karlH]

The latest on github is

https://github.com/karlheyes/icecast-kh/commit/8d7e04602e5cf6061e051d73148405b9e59e4f40

which updates the build setup, the actual update for the specific fix is a little before that, just look through the list on the site. So yes, after getting the log and you explaining the issue at hand I was able to identify and fix the issue.

The issue didn't affect listeners on sources, only if the listeners were on a fallback file and were pulling them through a cascaded fallback.

karl.

[square_eyes]

Great news. Thanks as always Karl.

I'll get the hosting provider to apply the latest patch and let you know how I go.

In the interim I have set up my own staging server 2.3.2 Win seems to handle the cascade just fine and simply relaying /stream to the hosting co. seems to follow along nicely.

[square_eyes]

Not an ideal response.

---
Hi, Sorry for the delay. We've been trying to research the patch. The administrator has said we will not be able to perform the patch at this time becasue we are not sure what it will do to the other clients on the server.
---

Oh well, I hope this has helped someone else.