FEED PROVIDER Issues

[SCPD]

On a Busy night (Friday or Saturday) I can't log in for several hours because the server is so hammered. In Addition, Ice Cast seems to drop an awful lot. I've seen Ice Cast tonight disconnect 4 times on my feed alone. Does LAPD or the other Bigger feed have this problem?

I would guess a Feed Provider should have access about anytime to adjust or edit a feed or content. But I can't get in sometimes for Several hours on Weekend nights.

So what's the issue with the Ice Cast drop outs? Lag? too many people? is My one to two occupant feed too much for the server?

This isn't worth the headache or the premium membership if I can't log in or experience continuous Ice Cast disconnects.

What's the deal?????

 

[kendrik578]

Winter time is typically RadioReference busiest time, so it could be so many people on the feeds causing it to drop. Believe me you aren't the only one who has this problem. My feed gets dropped just about everyday.

 

[blantonl]

If your feed is dropping, it is most likely due to your internet connection.

Our servers are not "hammered" and we aren't experiencing performance problems on the Web side.

MaxTracker, we always seem to have a consistent set of complaining from you - but we've always been able to determine that your internet connection (rural internet connection) is the most likely source of the problem.

 

[Chrome69]

Lindsay, you keep saying its our Connection, maybe its time you look for another server host to try if its your current one that you use.... Mine also drops daily as well only for seconds then reconnects. Just enough to drop the listeners and piss them off...

Try one server on another host company Lindsay and see if you have the same results or if its your current Server host... Most company's will give you a 30 day money back guarantee so nothing to lose and move some of the feeds that have this issue (like mine) over to it to experiment...

 

[SCPD]

Wierd my connection is not dropping out at all that i am aware of.

 

[hvscan]

Wirelessly posted (BlackBerry9530/5.0.0.328 Profile/MIDP-2.1 Configuration/CLDC-1.1 VendorID/105)

Same for mine - no known issues.

 

[Tweekerbob]

I'll echo that. On rare occasion do I see any of the RR servers (audio, web, etc.) "backed up". It can happen, but on very rare instances. For the most part, even my phone can pick up the feeds reliably as I travel through rural regions, eliminating the fact there is an issue with the server(s) as a whole.

I don't think the # of listeners has anything to do with it. If you understand the way that servers work, you'll get it. I've have 600+ listeners during a fire and no one was dropped. I'm sure others have seen such spikes...the servers continue to provide the feeds as if there was only one listener; no issues.

Overall, RR has provided a very reliable system in which people can provide feeds and likewise where people can listen to them. It's electronic. Can anyone show me an electronic system that is perfect (at least reasonably obtained)?

If anything is wrong, check your own set-up first. From there, let's investigate with very detailed info.

 

[matthewtomek]

I've never had any problems with my feeds dropping either. The only time there's been a problem, it's been on my end, not the server end. The max number I've ever had was 100 listeners, but everything went fine when the numbers got up that high.

 

[nexus]

Icecast use to drop my connection all the time, and some times would just lock up and time out. I changed to SCANNERCAST which is provided on here and it's been 100% stable for the last 3 or 4 months for the first time ever.

I suggest using scannercast over anything if you can help it.

 

[KMA367]

I've never had any problems with my feeds dropping either. The only time there's been a problem, it's been on my end, not the server end. The max number I've ever had was 100 listeners, but everything went fine when the numbers got up that high.

I'm still using Oddcast, and in the aftermath of this evening's Humboldt County earthquake my feed was handling over 1,200 listeners with no hiccups that I could tell. On a 3-year-old barebones Dell Inspiron box with WinXP.

My typical load is from about 2 to 10 listeners, so it was pretty heady stuff to have my rural little county's feed out-drawing the other 49 top feeds combined. Hearing occasional "gas leak" fire dispatches seven hours after the shake, some power outages, and alarm and sprinkler system problems and still holding 200+ listeners.

I've intermittently tried using Scannercast, but RealPlayer doesn't seem to get along with it yet.

 

[blantonl]

Lindsay, you keep saying its our Connection, maybe its time you look for another server host to try if its your current one that you use.... Mine also drops daily as well only for seconds then reconnects. Just enough to drop the listeners and piss them off...

Try one server on another host company Lindsay and see if you have the same results or if its your current Server host... Most company's will give you a 30 day money back guarantee so nothing to lose and move some of the feeds that have this issue (like mine) over to it to experiment...

Brian,

We're using Amazon EC2 with over 17 servers now. "trying" a new Web host isn't like copying a few HTML files over and giving it a 30 day try out. We spent almost 6 months last year evaluating and preparing for our infrastructure upgrades. I'm not trying to get on a high horse here, but listen man, we know what we're doing here and trust me that RadioReference isn't a single server-upload-your-stuff-via-ftp and away you go. This isn't amateur hour hobby time, this is a major web business operation.

Just so you understand the complexity and size of RR.... Last months stats:

* 3700th largest Web site in the *world* by web traffic
* served 32TB of outgoing bandwidth.
* provisioned and used 8TB of storage for NFS, backups, database, and audio archives
* Max concurrent sessions served: 3200
* 14 million recorded page views
* 760,000 unique visitors

Technologies used:

* Apache
* PHP
* Mysql (master, slave, backup)
* haproxy
* icecast / streamripper
* Amazon EC2, Amazon EBS, Amazon S3, Amazon Cloudfront, Amazon Elastic IPs,
* Vbulletin, MediaWiki, Photopost, Vbulletin Mobile
* SOAP / XML / JSON web services
* NFS / hyperic systems management
* ..and a zillion other things

In addition, we had over 1300+ listeners on one single feed last night - and that feed stayed up an running during an earthquake (I believe it went down for a power outage during the actual earthquake but came up when power was restored).

Folks, listen, last mile broadband internet connections aren't designed for 24 hour a day 7 day a week uninterrupted connectivity. Unless you have a business class network connection into your home (T-1, leased line, SDSL etc) you are going to face DSL modems resyncing and cable modems and network congestion, power blips, computer issues. 99% of the time the issue is NOT with our infrastructure. I'm not taking an arrogant position here, and I'm not saying that our infrastructure is perfect, but I'm telling you from experience that unless we're reporting an outage or problem on our end, it is most likely on your end.

 

[PeterGV]

I regularly struggle to explain this to folks who aren't network savvy: Forwarding a real-time, constant, stream of data from your home computer over the internet to a server is a very complex, tricky, thing. You need to expect issues. In fact, it's a wonder that it works at all.

Your feed software (such as ScannerCast, Edcast/Oddcast, or whatever) forwards at least 2 packets every second from your system to the RR server. Every packet must arrive, and it must arrive in a timely manner, or your connection will "drop."

If your connecting your computer to your home network with a wireless connection you're just asking for problems. Wireless is fine for occasional use, but you're just plain lucky if you can maintain a 24x7 connection over wireless. If you're sending your feed over wireless, one of the best things you can do is get a hard-wired connection.

Leaving your home network, if you're on a cable modem: Your internet connection bandwidth is shared with all your neighbors. If just one of your neighbors fires-up a download from NetFlix, your connection speed will be affected.

If you're on a DSL connection: You connection bandwidth is also shared, but it's shared at your Internet Service Provider's "hub" (the POP). It always makes me giggle when I hear small phone companies offer 20Mbps DSL to their customers... when the whole service provider might only have a total of 50Mbps or 100Mbps connection to the rest of the internet.

Problems aren't restricted to your home and your ISP, either. It gets worse as you go along the internet. There are no fewer than 12 "hops" between my house and the RR servers. That means there are 12 individual systems that need to handle every packet I forward to RR in a timely manner. If just one of those systems gets busy, bogged down, or one of THEIR connections experiences congestion, my connection can be dropped.

Look... your cable or phone company probably makes your internet connection sound terrific. And when surfing the web or downloading an occasional movie, it probably works great, too. But home internet connections aren't really designed to support a constant flow of traffic. They work, but they also have "issues." Business connections, where they need dedicated, continuous, traffic to flow to/from the internet, are very different than home connections.

That's why my dedicated 3Mbps service at work (2 T1 lines) costs about $2,000 per month -- and my 50Mbps (more than 15 times as fast) Comcast (cable modem) connection costs only $125 per month (and that's supposedly for "business" grade service).

You see... there's a lot to it, really.

Peter
K1PGV

 

[GTR8000]

This thread and threads like it amuse me to no end. Great posts by Lindsay and Peter explaining the enormous complexities of the "behind the scenes" workings.

RadioReference isn't a single server-upload-your-stuff-via-ftp and away you go. This isn't amateur hour hobby time

That quote had me in stitches, because it's so very true not just for RR, but for many operations like this. I'm always amazed by how many people are out there who think that because they have a domain with GoDaddy, they're all of a sudden an "expert" on all things Internet and network. Back in the day, I used to call these types "AOL experts", but I reckon that reference is pretty dated at this point. :twisted:

For the record, my stream runs on ScannerCast 24/7 over a modest residential cable service using an old school Linksys router that only has a half-duplex 10 Mbps WAN connection, and it STILL has an average uptime of 99% or better. I can be quite heavy with the uploading as well as downloading on this connection, and it still doesn't affect the stream negatively, even with that half-duplex WAN port causing all sorts of packet collisions. The only times it goes down are when: I take it down to reboot the computer; I take it down to update ScannerCast versions; my power goes out for longer than the UPS capacity; the cable connection drops (almost never); and last and least, there is RARE and very occasional glitch with ScannerCast or the RR servers.

If your stream is dropping regularly, it's a problem on YOUR end, not RR's. That's the bottom line here.

 

[NYRHKY94]

Great post Lindsay & Peter! I'm no technical guru like many of the folks on here, but I do see the exact latency Peter is talking about on my home cable modem service when the teens in my subdivision are home playing/downloading etc. The pipe is shared and I see & hear the difference.

Like you said, "most times" the service is just fine for the $$$ I pay each month for "consumer bandwidth". All in all, as a long-time user & feed provider on SA and now RR.....the uptime reliability has been very, very good. Between your post & Lindsay's, I think you guys have nailed it when it comes to setting the proper expectations for users & feed providers a like.

Keep up the good work!

 

[KMA367]

In addition, we had over 1300+ listeners on one single feed last night - and that feed stayed up an running during an earthquake (I believe it went down for a power outage during the actual earthquake but came up when power was restored).

The problem was indeed at my end of the feed. To be precise, our power didn't go out, but our internet service was out for five minutes immediately after the thing hit. As soon as my cable service returned, my Oddcast automatically reconnected (I wasn't home at the time) and it's been running absolutely non-stop for the 22 hours since.

You're doing a whole lot of something's right, Lindsay & company!

 

[Citywide173]

I did some checking on my feeds, and since RR relays from my shoutcast servers, not a direct feed from my computer to them, there is a much more convoluted path to get from my computer to RR. There are 9 hops to get from my house in southastern Massachusetts to my Shoutcast servers in Chicago, and I would assume at least 9 more to get to the Amazon servers in Seattle. With his in mind, you'd expect more dropouts than usual, but the RR feed has better than 99% uptime over the past two weeks (not counting the server issues-which RR announced).

I had noticed a reconnection issue that was occurring, but upgrading to the latest release of Proscan fixed that. You might want to look at your software options and see if there's one that stays connected longer in your application. I have to agree with everyone else above, the problem is definitely on your end.