My EMPEG serial number is 090000611. At home and at work, I have the same setup. Only difference is the PC that I run EMPLODE on. At work, I can sync 650MB of mp3s without any problems. At home, I get through about 15 or less songs and it gives me a synch error at stage 5, 0xffffffff Operation not permitted.

Both PCs run Win2k Server with latest patches, etc. Work has 128MB RAM, Home has 256MB. Work has a dual XEON 550 CPU, Home has AMD TBird 800. I use the same model hub (DLINK 10/100 auto switch) to connect everything. Work has a 3Com FastEtherlink, Home has a DLINK 530TX and both are 10/100. No other apps are running on Home but at Work I run Outlook, MSIE, Explorer, VC++, AOL IM, etc. and have no problems whatsoever.

This happened in 1.01.

Now, EMPEG has 1.02 loaded. and EMPLODE is 1.02. I did get a message from Mike Crowe about whether this was fixed in 1.02 and I just recently replied back saying it was not.

Just wondering if anyone else has problems like this? USB works fine but I would prefer ethernet as I have two PCs and I just want to leave my EMPEG in its rack when I am home and not move it to whatever PC I am working on. My EMPEG has its place in my HTS which is downstairs and my PCs are upstairs. Distance is not an issue as it was all wired by me and I did all manner of tests (continuity, bitrate, etc). I can easily connect my laptop downstairs instead of EMPEG and have no trouble going out to the Internet, etc.

Squid2k1
---
18GB of Green Mk II Fury - Fast As A Shark

Distance is not an issue as it was all wired by me and I did all manner of tests (continuity, bitrate, etc).

Did you use a genuine category-5 cable tester? Or did you just run some PC programs?

The fact that you wired it yourself actually makes me suspicious. Not to say that your wiring was bad or anything. But cat-5 cabling is pretty serious stuff, and there are some subtleties about wiring it that most laymen don't know. For instance, if you don't get the right wire pairs on the right pins, you can induce crosstalk which causes packet loss under heavy load. Which brings us to...

I can easily connect my laptop downstairs instead of EMPEG and have no trouble going out to the Internet, etc.

Yeah, but surfing the internet doesn't saturate the line, does it? An Empeg synch is probably the hardest work that wire is ever going to do. It's possible that the wiring could be the problem, and you just never "pushed" it until the Empeg came along.

Have you tested it with a shorter cable run? For instance, plugging it into the PC directly with a store-bought (not made-yourself) crossover cable? It's time to start eliminating failure points in the connection, and that's the first place to start.

___________
Tony Fabris

Did you use a genuine category-5 cable tester?
A genuine CAT 5 cable tester. Who ever thought my days of being a cable monkey would let me do fun stuff like wire up my EMPEG??

Yeah, but surfing the internet doesn't saturate the line, does it?
I am not surfing. I am saturating the hub, getting up to 50% utilization on it.

Have you tested it with a shorter cable run? For instance, plugging it into the PC directly with a store-bought (not made-yourself) crossover cable? It's time to start eliminating failure points in the connection, and that's the first place to start
I have tried it with store bought cable. I have tried it with a small cable with my EMPEG right next to my PC and a small hub between the two.

Trust me, I have tried all manner of tests. Nothing has yet to point to faulty wiring.

Squid2k1
---
18GB of Green Mk II Fury - Fast As A Shark

If you have access to a linux box, try pulling a sniffer on it, looking for dropped packets or mismatches. Bad ethernet card/driver possiblity.

I noticed some wierd stuff talking to the empeg over a transparent firewall bridge (don't ask... I just needed an ip addressable machine to play with, and guess what happened to be around.) Turned out to have a marginal ethernet card in the bridge, giving the errors. Doing a TCPDUMP on the connection was pretty interesting..

Trust me, I have tried all manner of tests. Nothing has yet to point to faulty wiring.

Okay, just checking.

___________
Tony Fabris

Yeah, I agree with synergy's assessment. At this point, you might want to be looking for card/driver problems. Sure, you can saturate the hub with internet applications, but I'll bet they're running error-correcting protocols that are more solid and more tolerant of lost packets.

(Agreed that if this is the case, then Emplode should be more tolerant as well, but I'm just trying to solve your immediate problem here.)

___________
Tony Fabris

Actually, wait. There's one other thing.

I went back and re-read the original message. You say the home unit is running an AMD CPU? Did the Empeg guys know about this when you initially reported the bug? I dunno if it's a factor or not, but it might be worth investigating.

___________
Tony Fabris

But cat-5 cabling is pretty serious stuff, and there are some subtleties about wiring it that most laymen don't know.

Christ it's only running at 10Mbit. Cat5 is overkill.

Sure, you can saturate the hub with internet
applications, but I'll bet they're running error-correcting protocols that are
more solid and more tolerant of lost packets.

Doesn't the mark2 over ethernet use TCP packets?

Doesn't the mark2 over ethernet use TCP packets?

Yes, it does. Bad network cabling is IMO not the problem here. It's more likely that your home network (with its lack of collisions compared to the office one) has too high a throughput for the empeg to cope -- although 1.02 should be able to cope even with full wire speed transfer.

Peter

Greetings!

Is there any documentation on what the stages are for synchronization and what the errors are?

Paul G.
SN# 090000587 (40GB Green)

Unfortunately, no access to a linux box but I guess i could use NetXray on my PC and capture the packets...I will try this on the weekend.

Squid2k1
---
18GB of Green Mk II Fury - Fast As A Shark

You say the home unit is running an AMD CPU? Did the Empeg guys know about this when you initially reported the bug? I dunno if it's a factor or not, but it might be worth investigating
Yes I pasted the same message that I sent to bugs@ in the forum here so I assume they saw and read/understood it...

Squid2k1
---
18GB of Green Mk II Fury - Fast As A Shark

It's more likely that your home network (with its lack of collisions compared to the office one) has too high a throughput for the empeg to cope
But I have no control over how fast emplode is pumping data...do I? My PC may be doing 100Mb/s and the EMPEG only does 10Mb/s but I think that if there are collisions TCP takes care of all the resyn acking, etc. I have been on the office network when no one is around so traffic is very miniscule and I have had no problems with the synching of the EMPEG.

Thanks for all the input!

Squid2k1
---
18GB of Green Mk II Fury - Fast As A Shark

Okay, just checking
I appreciate it. I always make sure the user is not at fault first!!

Squid2k1
---
18GB of Green Mk II Fury - Fast As A Shark

No.

This area of emplode is greatly improved in v1.1 - the stages are named, not numbered, and we're investigating using descriptions instead of numbers for the errors.

Hopefully, this'll make it easier to figure out what's gone wrong during a sync.



Roger - not necessarily speaking for empeg

Yes, it does. Bad network cabling is IMO not the problem here. It's more likely that your home network (with its lack of collisions compared to the office one) has too high a throughput for the empeg to cope -- although 1.02 should be able to cope even with full wire speed transfer.


I agree it's probably not the cabling... Cat 3 would do fine in this situation, but I also disagree here. He's using a switch which should eliminate the collision difference. Furthermore, I've pumping into EMPEG from a gigabit network card at work, into a HP switch and then throttled down to 10Mbs to the EMPEG. Worst case situation, the switch would slow everyone down to 10mbs, but I don't think the even the DLINK $65 4 port switches are that dumb.

The biggest thing I've seen is a sensitivity to timing errors on the EMPEG protocol. In the very few tests I've ran on the network spec itself, if you delay a packet or cause a retransmission, emplode seems to hang for a bit. I've only been able to fuxor it up enough to actually cause an error a few times however.

And as an aside on this, it could very well be the actual network card in the PC. Depending on which version of the 530tx he has, it's either the Tulip chipset or the RTL8139. I have had AWFUL luck with the Tulip variant of this card in Tbirds (went through 4 of them... all of which worked fine in BX chipsets, but failed weirdly in the VIA.)

That's the biggest reason I'd say put a sniffer of some sort on the network and look to see if there are short, stub, or mangled packets. Assuming the cabling and switch is good, there should be none at all. Given that his empeg syncs at work, I'd say it was ok.

Hell, Linksys cards at $15 at compusa, you could always use a spare.

Depending on which version of the 530tx he has...

IIRC, if it's a late model DFE-530TX, then it could be a Via-Rhine 2 or an RTL8139.

I've got one at home (the ViaR2 flavour), but I haven't installed it, due to severe lack of driver support - it's not in stock Linux 2.2, and W2K didn't recognise it.

I've not, on the other hand, ever had any trouble with my RTL8139 (or is it a -29? - I don't recall) or a Samsung-badged Tulip card.

It's been so long since I took my empeg home, I don't remember whether I've tried it on my home PC (which is an Asus A7V w/T900) - I'll take it home tonight, and maybe give it a try.

BTW, the Tulip is in another A7V-T900 (running Linux), and I've had no trouble with it. Although I've not tried it with the empeg recently.



Roger - not necessarily speaking for empeg

RE: Tulip cards..

I purchased about 30 of them at a time, had trouble and moved on. I know that some tulips work correctly and well, but it seems (RE: Donald Becker, I believe) that DEC is changing the internal core of the tulip chip, while keeping the interface pretty much compatible. Some of them work better than others compatibility wise. In my case, I knew almost immediately that the card wasn't working correctly, from the interface statistics on my switches... Not to mention shitty throughput and latency.

RE: Via-rhine...

I didn't realize that the 530TX was hiting the Rhines as well. Does anyone remember the days when manufacturers actually changed the part number when the part changed???????