MRV Issue Between 9.3 and 9.4 Units

[TiVoDoughBoy]

I have two TiVo HD units, one that's received the 9.4 update and one still at the previously most current 9.3 release. I need to do some additional testing, but tonight I went to do a transfer from the 9.4 box to the 9.3 and the transfer time was painfully slow, less than real-time.

When they both went to 9.3, MRV performance drastically improved -- a 60-minute SD show transferred in 4 or 5 minutes and an HD show in about 20 minutes. IIRC, when I did the math it was about a consistent 32- or 33-Mbps transfer rate.

Both units are wired to 100-Mbps switches and there was no other network activity at the time, and when I did a transfer tonight of "Kitchen Nightmares" (in HD, and again from 9.4 to 9.3), it took somewhere around 2 hours (rough estimate) to do the transfer -- or about 6-Mbps.

Again, I want to do some more testing such as trying another HD program in the same "direction" (9.4 to 9.3), trying an SD recording, and also going from 9.3 to 9.4, but I wonder if anyone else is in a similar situation, i.e. two units, one that's received the update and one that hasn't, and has noticed the same issue?

My biggest concern is that this is indicative of a problem that's going to exist at 9.4 or if it's just a temporary issue while I have two different versions running. And if it's only temporary, if it's possible to call Customer Service and request that the update be pushed to my other unit ASAP -- I know it sounds whiny, but we use MRV a lot at our house and in the current situation, it's not usable, a "service disruption" as far as I'm concerned...

[TexasGrillChef]

Have you called TiVo yet to see if they can push it?

TGC

[TiVoDoughBoy]

Have you called TiVo yet to see if they can push it?

Not yet. Want to test it out some more first; just wondering if anyone else has seen the issue.

[dbenrosen]

I didn't check as closely as you did, but my HD transfer seemed to be pretty quick from the 9.4 to 9.3 boxes. By quick I mean I started watching right after starting the transfer and didn't run into any pauses even after fast-forwarding through commercials.

[TiVoDoughBoy]

Want to test it out some more first...

More results... saw the same dreadfully slow performance with a different HD program and going in the opposite direction (9.3 to 9.4).

[doconeill]

I suggest you try:

- Disconnect and reconnect all network connections from the switches.
- Power cycle the switches

It is possible one or more device connections came up in half-duplex instead of full-duplex, which during high network load would result in poor actual throughput.

[TiVoDoughBoy]

It is possible one or more device connections came up in half-duplex instead of full-duplex, which during high network load would result in poor actual throughput.

Believe it or not, that was the first thing I considered and have already reset everything (I'm an IT guy :))... figured I'd leave out that detail instead of further confusing the problem, but good thinking! Anyway, same issue. Grrr...

[TiVoDoughBoy]

Time for a little self-deprecation...

(I'm an IT guy :))...

Translation = (I think I know everything :))

So mea culpa here, about a week-and-a-half ago, I bought a SlingBox before heading out of town and put it on the TiVo HD unit in our bedroom (less likely to interrupt someone's viewing at home during the day). For cabling reasons (and laziness), I put another small four-port 10/100 switch right at the location where the TiVo and SlingBox now are and slid it under the dresser: out of sight = out of mind, right?

So when I checked all my connections to make sure they were connecting at not only full-duplex, but also 100-Mbps, I found that on that switch, none of the three connections (SlingBox, TiVo, and uplink to the jack across the room) were connected at 100/full. I reset it, and voila, all is well again.

So, to dconneill,
It is possible one or more device connections came up in half-duplex instead of full-duplex, which during high network load would result in poor actual throughput.

thanks for the reminder. The scary part in all of this? A number of years back when I was working on my Master's I did an Independent Study course -- the topic of the research paper? Ethernet autonegotiation. No joke. *sigh*

Going now to make sure my "defective PC" is plugged into a power outlet.

[moyekj]

So when I checked all my connections to make sure they were connecting at not only full-duplex, but also 100-Mbps, I found that on that switch, none of the three connections (SlingBox, TiVo, and uplink to the jack across the room) were connected at 100/full. I reset it, and voila, all is well again. Perhaps others can benefit from this too. Exactly how do you determine for sure devices such as SlingBox, Tivo, etc. are running at full duplex 100Mbps on a switch or router?

[lrhorer]

I have two TiVo HD units, one that's received the 9.4 update and one still at the previously most current 9.3 release.
I definitely have not seen that. Just tonight I downloaded a 1 hour 1080i program from my 9.4 S3 to my 9.3 S3, and it took less than 30 minutes. As you can see below, the transfer rate to the 9.3 S3 was 34.87Mb/s, which is respectable, although not astounding. The transfer a few minutes ago from my video server using pyTivo as you can see came in at 15.91 Mb/s. The Video Download was Cranky Geeks. Of course, the fact mine are S3s and yours are THDs could make a difference.

Both units are wired to 100-Mbps switches and there was no other network activity at the time
Did both by any chance have both of their inputs tuned to HD channels? This can definitey slow down a transfer, although I would not think as much as you report.

http://fletchergeek.com/images/SN850039.JPG

Edit: Oops! I just saw your mea culpa. I should have read further before posting.

[lrhorer]

Perhaps others can benefit from this too. Exactly how do you determine for sure devices such as SlingBox, Tivo, etc. are running at full duplex 100Mbps on a switch or router?
It's not always easy. Managed switches help. Of course if the device has an indicator or can be logged into, one may merely check whatever utility affords. Sniffing the network can show the presence of runts, checksum errors, invalid frames, jerky packet flow, and lots of re-transmissions. One clue can be if a large file transfer starts out at very high rates but immediately drops down to very low rates. Also, if all files smaller than the default TCP Rx buffer (typically 8K or 16K) positively zip right across but files of only a few hundred K dawdle (comparatively speaking), a duplex mismatch might be the culprit. If there is no duplex mismatch, but instead both ends of a link are half duplex instead of full, then network performance will be a bit less but still decent for transfers in one direction, but transferring two files simultaneously in opposite directions will cause the transfer rate to drop precipitously. If both ends are properly full duplex, then transfers of two files in opposite directions will only be slightly slower than transferring one file, provided the hard drives on both sides can keep up. This assumes the latency betwen the two systems is small.

[doconeill]

thanks for the reminder. The scary part in all of this? A number of years back when I was working on my Master's I did an Independent Study course -- the topic of the research paper? Ethernet autonegotiation. No joke. *sigh*

Then you should know better than to trust it... :)

On switches where we can set duplex (usually higher-end managed switches), we always force duplex because we find that there is a non-zero chance that it will negotiate wrong, and people start complaining why the server is so slow...

Most home switches cannot be forced, however. Some will at least have an indicator letting you know, but if you can't get status or at least statistics from either end, its basically a "plug it in an hope for the best".

[TiVoDoughBoy]

Perhaps others can benefit from this too. Exactly how do you determine for sure devices such as SlingBox, Tivo, etc. are running at full duplex 100Mbps on a switch or router?

Everyone else pretty much covered it, but in my case, I have a number of small 4- or 8-port Linksys switches around the house in locations where I have "density" of several Ethernet devices (TiVos, Slingbox, PCs, printers, IP telephones, etc. -- and unless I have something I specifically want to be mobile around the house, I prefer to wire it). All of the Linksys switches I use have indicator lights that have been generally trustworthy. But the indicator lights do *not* show duplex (half or full), only speed (10 or 100). With this "problem", though, the indicator for all the devices for running at 100 was unlit, and since I knew everything plugged in was 100-Mbps capable, I knew something was up.

dconeill is correct -- unfortunately with lower-end, unmanaged devices, you can't force speed or duplex, it's set to autonegotiate by default. The TiVos and SlingBoxes are the same, you can't force them either. However, if you do come across a device that will let you force it and plug it into an unmanaged switch, *don't* force it on the device. Reason being is that for autonegotiate to work, both sides need to be set to autonegotiate. If one side is set to 100/full and the other is set to autonegotiate, the autonegotiate side will "negotiate" to 100/half and you'll have a duplex mismatch (because forcing one side causes it to not "advertise" its capabilities to the other side).

[lrhorer]

Everyone else pretty much covered it, but in my case, I have a number of small 4- or 8-port Linksys switches around the house in locations where I have "density" of several Ethernet devices (TiVos, Slingbox, PCs, printers, IP telephones, etc.
Yeah, me, too. My main switch is a TrendNet 24 port managed 10/100/1000 BaseT switch. I really like it, especially given the price, which was quite reasonable for a 24 port managed Gig switch. The house wiring distributes 4 pair Cat5e all over the house from the server room, but at several of the drop points I have more than one Ethernet device, so I put little 5 port dumb Gig switches there. I don't buy switches that don't have some sort of duplex indication, and switches with LED duplex indicators are only a few dollars more than ones which don't. It hasn't happened often, but the one or two times the inidcators have prevented me from chasing a red herring have more than made up for the tiny cost difference.

[pcbrew]

I have never seen the Transfer History screen shown above.
How to you get to it?
Is it new in 9.4?

[moyekj]

I have never seen the Transfer History screen shown above.
How to you get to it?
Is it new in 9.4? Yes, new in 9.4.

[TiVoDoughBoy]

I have never seen the Transfer History screen shown above.
How to you get to it?
Is it new in 9.4?

It's under Messages & Settings --> Settings --> Phone & Network --> View network diagnostics --> Transfer history

And yes, new in 9.4.

[web1b]

It's under Messages & Settings --> Settings --> Phone & Network --> View network diagnostics --> Transfer history

And yes, new in 9.4.

I just looked there and under "Videos copied from remote device" and "Videos copied to remote device" is says "None" even though I transfer many video podcasts from iTunes using Tivo Desktop almost daily.
On Video download it says 2.13 Mb/s and YouTube Video it says 1.37 Mb/s

I have a high speed cable internet connection that is more like 8 Mb/s and the internal network is wireless 802.11N using the Tivo's 802.11G adapter.

Is there something wrong why it lists the very slow video download speeds and doesn't even list the videos transferred from the PC at all?
I think my video transfer speed really is slow because I usually have to wait for almost the entire video to be transferred from the PC before the Tivo will allow it to be played and this can take several hours if it is an hour long HD video.