servers not clustering

[stormwatchradio]

Wondering if there are special settings required to ensure that all relays cluster properly. I have noticed that sometimes when the yp is slow that this has been a problem and I am guessing that yp is the one who does the cluster as all the servers are working fine and taking listners they just wont cluster.

We are using a server A as a main souce 1 connection in RIP mode
server B has 350 connections and relays from server A and is set to public=always so that it wont go private

servers C and D connect to server B and have 50 and 100 connections respectively.

obviously we cannot serve all 3 off of the A server as that would be 3 connections up at 64k but we dont like sending to a server offisite we prefer that it relays off of a server running on our local machine as we have found that this limits skipping and lag etc.

all 3 servers

Any suggestions?

thanks

[FesterHead]

What is the station name?

Sometimes it takes a couple minutes for the relays to cluster up.

[stormwatchradio]

oh we realise this they havent clustered since 1 am last nite http://www.shoutcast.com/directory/?...x&numresult=25

[DJHotIce]

at the time of this post servers *Do Not* Appear clustered.

BTW Have you allowed AllowRelay=Yes in your config?

[stormwatchradio]

yes they are allow relay well 2 of them relay off of the 3rd which has allow relay yes as do the other 2 and all are set to public=always so they should be clustered it doesnt make any sense.

[FesterHead]

Is the source dropping/reconnecting often?

[stormwatchradio]

nope everythings fine they have been online since 1 am eastern last nite and have handled hundreds today fine

[stormwatchradio]

one of the servers the 50 user one had some issues yesterday near them as a router or gateway must have been acting up so we decided to make the 350 user the main relay and the other 2 relay off of them and did this last nite and ever since they havent clustered i have had this happen before a while ago and they did cluster eventually after a few stops and starts i just hate doing this when i have 70 some people still on and hate doing it all all hehe just wanted to make sure that there wasnt a setting maybe the yp was down for a few when we brought them back up last nite

[FesterHead]

Do they cluster if you run DNAS A in non-RIP, public=default mode (make sure to reserve the IP of the DNAS relay that connects to it)?

[stormwatchradio]

the DNAS with the 1 connection locally on the streaming machine has allways been in RIP ONLY and this has never been an issue i will map out below what order the servers were in before and what order they are in now

***before**

1/1 main source RIP Only
50 users server
100 user and 350 user connected to the 50 user server

***now***


1/1 main source RIP Only
350 user
100 user and 50 user connected to the 350 user server

no settings changed at all on any config other then sending to different relays

[FesterHead]

Please examine the DNAS log files for dropped source connections.

[stormwatchradio]

<12/09/03@21:59:20> [main] Client Stream thread [0] starting
<12/09/03@21:59:20> [main] client main thread starting
<12/09/03@22:00:13> [source] connected from 127.0.0.1
<12/09/03@22:00:13> [source] icy-name:::OldiesMix Radio:: The Internet's New Mix of Oldies: playing the Greatest Hits from Yesteryear ; icy-genreldies country top 40 80s
<12/09/03@22:00:13> [source] icy-pub:0 ; icy-br:64 ; icy-url:http://www.oldiesmix.com
<12/09/03@22:00:13> [source] icy-irc; icy-icq:NA ; icy-aim:davyjca
<12/09/03@22:03:14> [dest: 69.93.**.***] Added to Reserve IP List
<12/09/03@22:03:38> [dest: 69.93.**.***] starting stream (UID: 5)[L: 1]{A: Shoutcast Server 1.9.2}(P: 0)
<12/10/03@00:36:54> [dest: 64.230.***.109] kicked
<12/10/03@00:36:54> [dest: 69.93.**.***] connection closed (9195 seconds) (UID: 5)[L: 0]{Bytes: 74634496}(P: 0)
<12/10/03@00:37:58> [dest: 69.93.**.***] starting stream (UID: 6)[L: 1]{A: Shoutcast Server 1.9.2}(P: 1)



so i kicked the 50 user and added a RIP then the 350 user connected last night and thats the end of the log so no source disconnects

[stormwatchradio]

well we just shut things off and restarted lets see what happens other then losing 50 listners hehe

[stormwatchradio]

ok it says cluster now on the 350 but nothings clustered with it yet they still show alone hopefully that clears up otherwise i cant think of anything else

[FesterHead]

I'm baffled.

[edit]

Try use public=Never instead of RIP mode.

(Make sure to still reserve the relay IP, though)

[/edit]

[stormwatchradio]

well the only thing i can say is in the past i have seen weird things like servers that have been offline for 48hrs still showing as backups hehe so maybe its cached somehow i dont know enough about the yp setup to say. all 3 servers are getting yp touches but they arent saying backup server msgs

[Tom]

If the one not clustering has more than one IP use the DestIP setting in sc_serv.ini/sc_serv.conf

Tom

[stormwatchradio]

OK worked like a charm thanks Tom didnt have to use the destination ip simply told the relays the 1st ip in the block. So correct me if I am wrong but my understanding is if you have a server on a box that has numerous ips attached to it then you either need to tell it which ip it is or you have to make sure that the relays use the 1st ip in the block so that the servers and yp don't get confused as they all think that it is that 1st ip and not a secondary one. Just thought id type that out for others who learn from this type of thing and can refer back to it.

[Tom]

As long as the IP the relays use to relay the main server match the IP that the main server contacts the SHOUTcast yp on then they should cluster. As for how you want to accomplish that well that is up to you.

Tom