Bitcoin Forum

Hi, I've been struggling for over a week with my blades. I've found one other person with the same problem I have, but their work around didn't help me. I'm very frustrated with this at this point, and I just want to get them to work properly, and I'll gladly pay anyone that can offer me a proper solution, and I greatly appreciate any help that anyone can offer at this point.

Here's the problem. The blades connect via getwork to any pool. They hash between 4.5GH/s - 5.5GH/s at any getwork pool. They will connect to the stratum proxy and receive/submit work for a few seconds. This usually results in 8-40 received units of work and 0-4 submitted units of work. Then they stop. The blades will then reset after about two minutes because they're not hashing. Now, if I disable the wireless radio on my Verizon MI424WR Rev. E router then the blades will connect to the stratum proxy just fine, and hash ~12.5GH/s. They perform better via getwork as well, around 10GH/s. What the other person did to get around this was to connect a different wireless router and use that for wireless. I tried this, and still have the same problem - the blades will only perform properly with the connected wireless router is disabled. Leaving the wireless disabled isn't an option for me.

The stratum proxy works perfectly with other mining devices. I've tried the stratum proxy on multiple local computers with the same results. The blades connect to a switch that connects to the Verizon router. I have bypassed the switch, running one blade directly from the router with the same results. I have set up QoS to prioritize all traffic to and from the blades with no change in results. I've tried modifying any wireless setting I could think of, but nothing makes a difference other than shutting the wireless off. Beyond the fact that I really want to get these blades working, I'm really curious to know what could possibly be causing this to happen. Like I said, any help would be greatly appreciated. Thanks guys.

Is your stratum server connected to the verizon unit, or the switch that's attached to the blades?

Currently the switch that is connected to the blades.

How many blades you have now? If you have more than one, please confirm that the ethernet controller has no problem.

I remembered you host stratum proxy by Raspberry Pi. Are you also using the same Pi as the proxy for other mining devices?

I suggest you set a static ip address for the Pi if it is a dynamic ip.

I have two blades, swapped the Ethernet controllers with no change. I don't have a raspberry pi, but I have set a static ip for the computer with the stratum proxy.

Have you tried using different mining software?  I would try bitminter.com if you're currently using cgminer and see how high that can push the blades.

So it sounds like it's just RF interference? With the blades not being enclosed in any sort of housing they're pretty vulnerable to this I suppose. Have you tried physically relocating them to another room?

Try changing channels on your wifi router.

Have you tried lowering the transmit power on the wireless options?... should be under the advanced wireless options section of that verizon UI...

EDIT:  Do you have the model of the router that the other guy used for his Avalon issues?... might be worth just running out and trying to get that specific model router, disabling the wifi on your verizon router, and using the other router solely for wireless access.

EDIT #2:  I would also attempt to turn off the CTS protection under the advanced wireless options (if it is turned on)... and if it is already turned off, try turning it on. This may be an issue if you have a lot of wifi devices connected, but I would try it anyway.

Couple things to try. First, does your Verizon router allow you to adjust QOS or do port forwarding?

I would enable port forwarding, and forward the appropriate ports to the appropriate devices (are you using a local stratum server, or a stratum pool)?

Using QoS to give package traffic priority to your devices over any other devices might not be a bad idea either. As well as changing wireless channels, if your router allows.

If you are using another wireless router, you will probably have to configure the modem to pass EVERYTHING through to that router, meaning that your router is the one that authenticates to the Verizon Network with your credentials, and gets the external IP address (Ie; configure PPPoE on the replacement, and disable network authentication on the Verizon router/modem).

If your router's "external" or "WAN" IP starts with 10. or 192., you don't have an external IP. Call Verizon, and ask them for your network authentication (PPPoE) credentials, if you want to use another router to manage the network.

And, rather than do a 3rd edit on my post above -

To troubleshoot, I would look at the 'system monitor' in the UI... and look at your network statistics for your device(s)... see if there are any errors or dropped packets:

should look something like this: (random image I found)
http://blog.jeffreyberg.net/wp-content/uploads/2011/12/1_SystemWideMonitoring_Pre-e1323895181186.png

EDIT:  I can also use teamviewer to troubleshoot, if nothing above works.

i can remote in via team viewer.  msg me.

Ditto to the guys offering remote support.

Is it possible the wireless is giving out your blade's ip address to one of your wireless devices?  Usually you can limit the ip addresses given out by dhcp.  Make sure your blade's ip is outside of this range or specifically tell your router to reserve it to your blade's mac address.

lower MTU

The problem with offering a bounty like this is people are going to blanket spam you with often irrelevant things.

Try setting the difficulty to 16 in pool.

I'd expand that to don't remote with ANYONE, they're no more likely to be able to fix it looking at your desktop than they are on here.

At one point I thought it might just be interference too, I lowered the transmit power to 15% with no change. The blades are already pretty far from the wireless hub, about 100 feet. I could get them maybe another 15 ft. away, but I doubt it would make much of a difference.


I've tried every channel, no change.


CTS isn't an option on either of my routers. I'll probably look at getting that exact model.


I've poured over the logs for the router, no errors are showing up. No errors or dropped packets show up on the system monitor either.


The blades have static, reserved IP addresses from the router.


I know, I'd rather sort through a bunch of responses for something that might work than doing nothing though. I've set the difficulty in the pool to a wide range from 1 - 1000, it makes no difference.

You realize that it would be his actual ROUTER config that we would look at, and not just his "desktop" right? And yes, being able to actually LOOK at what's going on is much more useful than the shotgun approach of just shouting out your best guess, having none of the relevant information.  ::)

I'd make sure all hosts / adapters have the same subnet, same gateway, same dns server but different ip addresses than each other.


if your blades are static, they should have the dns and gateway set to use your routers ip address on the LAN.

you may have to enable dmz or port forwarding as well it really depends on how your network is setup.

without any remote connectivity I can't help you too much other than guess what your problem is.

if disabling the wifi somehow gets the connection to work, IP connectivity shouldn't have anything to do with it and it shouldn't have anything to do with DNS/IPs/Gateways/port forwarding/etc... and the fact that some of the hashrate gets registered using ethernet, it means the ip connectivity exists enough to transmit information to the pool, which means the ports/gateways/etc all should be fine. Unless I am not understanding the steps you took to troubleshoot this... I would assume its about the wifi either interfering with the frequency that the chips or some other part of the machine runs at... or that there are so many wireless devices that the CTS setting is required to avoid collisions, but I honestly have never seen this setting do much in other networks... could work for you though!  ;D

If it is a verizon router, you should have access to the cts setting (or you should, I know I see it on mine...) after going to 'wireless settings' at the top, go to 'advanced security settings' on the left, and then at the bottom click 'other advanced wireless options'

This gets you access to the transmit power/cts setting/etc... but its possible you have an older firmware or an older UI on your router... teamviewer might be best to troubleshoot this. (in which case, try updating the firmware under the 'advanced' menu)

Although, again, I would think getting the same exact model router as the other guy would fix the situation (if the issues are identical)
 

EDIT:  There is also a 'Wireless QOS' area that you could mess with as well... not sure if you modified the main QOS settings, or the wireless QOS settings, but on my verizon router they are separate.    (*The Wireless QOS settings are off of the same page with the CTS setting - at the bottom under the "WMM settings" link)

EDIT #2: (now that I have access to my verizon router - screenshot http://postimg.org/image/mjo7ac291/)

EDIT #3: modified due to my misunderstanding of the initial setup  :P

As I read it his blades are all wired to a switch which is wired to the router.  The wireless is just for his non-mining wireless devices, right?  What does it say on your router info page that lists all attached devices and their ip addresses?

I believe you are reading it correctly, I actually sent a PM earlier after my post to re-clarify this, because I am pretty sure I  misinterpreted the initial setup.  :P  The previous post still makes sense to me with a few word changes... and a strikethrough on the frequency comment, because I am not really sure this would be as much of an issue if the blades are wired to the switch (and are 100ft from the router)...



Understanding the setup now, I'd focus on the settings/config of the router... you've probably tried all of this, but changing the security type (switching to legacy mode, compatibility mode), disabling wep/wpa security entirely, disabling the Wireless QOS settings, etc... it sounds like you went through most of the wireless settings already. Can you provide the model # of the router that the other guy used?  there may be an obvious feature of that router that might shed some more light.

If I was you, I'd want to confirm 100% its not the blade. Take it to work, take it to a friend, hell take it to Starbucks. Just see if you can get 13 GH on another network. If you can, its your network and that's where we look.

If not, its back to friedcat as I already checked everything else for you.

Have you tried solo mining locally with wireless on?  Might help further isolate the issue.

Having similar problems as the OP, I'm going to attempt this this weekend.

Also, try upgrading firmware on router.

Try turning it on and off

Sorry -- couldn't resist

This is pretty much my assumption as well.


I appreciate the screenshot, I'm not sure how but I missed the 'other advanced wireless options' link at the bottom of that page until now. Seems like a strange place to put it based on how the rest of the UI is laid out. Unfortunately CTS (both cts-only and rts-cts) mode didn't make a difference. I tried lowering the RTS threshold in conjunction with this also without success.

I previously modified the main QOS page, which had a section for wireless. I don't see a wirless QOS section on the page with the CTS settings. I promise I looked thoroughly this time :)


This is correct. I have a few roommates who's PCs connect wirelessly, as well as cell phones, tablets, etc.


Firmware is up to date.


Haha, thanks for the laugh.

I ordered the other router, and had it shipped overnight so it'll be here today, the 24th, and hopefully that works around the issue and gets the blades up and running. Thanks for all the suggestions and input so far.

If you are using wifi for connectivity to blades and network.  Its very likely that you blades are getting interference from other wifi networks.   This will especially happen if you live in an apartment complex.

I suggest try using channel 1, 6, or 11  (assuming its 2.4ghz).  


If you have 5ghz spectrum use the first available channels.

5ghz channel has 20 non overlapping channels and will thus have less interference.

2.4ghz has only 3 non overlapping channels, and 150-300mbit/sec 2.4ghz n radios will use 2/3 of the spectrum alone.


It sounds like your options are:
 a.  use 802.11  5ghz spectrum
b.  get different antennas on your blades
c.  move your blades elsewhere
d.  run ethernet
e.  run power line ethernet

I'm not totally sure why you are using getwork on a blade as I don't own one.  But I guess its necessary to use it for the blade and then you use a stratum proxy to reduce your outside network bandwidth to the internet?

How fast is your verizon internet upload and download speeds?   what is its milliseconds in latency / ping time?

I've tried all wifi channels 1-11 with no change. The verizon router I have is b/g only, no n. The blades aren't hooked up wirelessly, they're connected via ethernet to a switch that connects via ethernet to the router. I'm using getwork because the ASICminer blades are only capable of getwork protocol directly. To get the most efficient use out of them you can use a stratum proxy to connect to a stratum enabled server and provide work to the blades.

I live in a house with a fairly large yard, no other wifi networks in range.


It was from this post:

https://bitcointalk.org/index.php?topic=208212.0

Wait.... do you think there is a max clients limitation somewhere? Too many phones connecting which is disrupting DCHP or other?

I've been reading about these routers and apparently they have a very small NAT table, which causes issues for people with many devices. However since restarting the router should clear the NAT table, and restarting the router doesn't make a difference with my problem, I ruled it out. Right now there's only 2 connected wireless devices, the two blades, and two PCs connected via ethernet, and there is no change in the problem with the blades.

It is not really called 'wireless qos' on that page, but is under a link called 'WMM Settings' at the bottom... the link is also in the screenshot I posted, its just a text link at the bottom of the page... the verizon interface is definitely a little convoluted!  Might just be something that is available on my verizon router but worth checking again if you were looking for a 'wireless qos section'  ::)

Hopefully the other router will be the quickest solution at this point!  

EDIT:  The model router I am looking at is a MI424WR-GEN2   (which has the WMM settings link)

Since wireless routers are cheap, just get another router or access point for your wireless needs, and turn the wireless off on that Verizon POS.

Check the heatsink on the blade. Re-apply thermal paste if needed. Cool front and back of blade with fans.

And this has WHAT to do with this guy's problem? Did you read his problem description? The issue is only replicated when he turns on the wireless radio on his router. Nothing to do with cooling. Zero. Zip. Zilch. Nada.

i would go this course of action suggested.  buy a $15 router on craigslist or I can sell you a really nice one for like $40 in btc that does 2.4ghz and 5ghz

Not sure some of the recent posts have been reading the entire thread  ;D

EDIT: ( at least, I hope this is the case... :o )

This has already been recommended... and this has already been tried. In the OP he said he already tried a different router and it didn't work. I had suggested to get the 'exact' model of the other user who had the same issue, then disable the wifi on the verizon router... he ordered the router and it should arrive today. (hopefully it works and this thread can be done with and we can all focus on bigger and better things  :D)

Do you have a second NIC lying around?  Since the blades only need to connect to your stratum proxy you might try creating a new LAN with a different subnet and hooking your blade switch directly to your proxy computer.  If you don't have a second NIC you could test by installing the proxy on a laptop and using its ethernet controller as the new lan and its wireless to connect to your router.

And I forget if you've tried it without the switch?  Connect one blade and your proxy directly to your router.

So you know what the problem is? Is this already fixed? Everything is worth a try in my opinion.

That's known as the "Shotgun approach" and it rarely works. There is a methodology to troubleshooting things. You don't rebuild the engine when the tires are wobbling.  ::)

I don't believe this will work, because the issue doesn't seem to have much to do with the connectivity between the blades and the proxy...  (because the traffic still has to pass through the verizon router to get to the internet, and I feel like the router is causing the issue)... although the laptop idea may be something to try... because that forces the traffic through the wifi which may have different results, might be the same issue though... it would be worth a try if you have a laptop or a wifi card you could put onto the proxy system.

But, this did give me another idea if the new router doesn't work... which would be to try and segment the wifi/wired traffic and basically create 2 different vlans on the router. This *may* force the router to handle the traffic differently and may circumvent the problem... but setting up vlans on a verizon router can prove challenging... and involves breaking the bridged connections that Verizon routers create, creating the vlans and assigning them to specific ethernet ports. I have done it before... it's just not pretty, especially if you ever have to reset that verizon router (which verizon support loves doing in the event of any issue)... here is a basic guide I found for doing it: http://blog.jeffreyberg.net/?p=56

But you need to be prepared to handle any troubleshooting once you start this process, as you may drop off the internet while doing it.

This will narrow it down to some weird wifi interference with the blades (which would be odd) and some ip address dhcp conflict, which seems more likely.  Unless one of the wireless devices is running upnp and is stealing port the stratum port (3333?) for itself.

I just got the new router, I'll be setting it up in a couple minutes. If it doesn't work I'll probably try setting up a couple vlans and seeing if segregating the traffic makes a difference. I really doubt its wifi interference, as I would imagine the amount of other people using blades and having a wireless network running simultaneously would be creating many more problems than my little posting. I'll let you guys know if this new router solves things.

Another idea would be to try and get the proxy system connected to the switch (if it isn't already)... it sounds like the proxy is currently connected to the router, which means the blades have to go through the router to communicate to the proxy, then the proxy comes back through the router and out to the pool... might be a simpler setup if the proxy is connect to the switch, and the blades communicate through the switch to the proxy system... and ultimately out through the router when communicating to the pool.

I am starting to stretch for ideas now, so I am going to wait until Nave replies to go any further... before my head explodes  :P  


EDIT:  and, of course you reply now... good luck with the new router, hopefully the easy solution!

I love the WRT54GL for a wireless router, nothing beats old school.

If the new router doesn't work... you may also need to try and find the same firmware version that the other guy is using - if it is older...
 (as a last ditch effort to try and get the new router to work)

New router connected, using verizon router as a bridge. Same problem. Ready to chuck blades out the window. Taking a break, be back later.

The proxy system is on the switch with the blades. I've tried running the proxy on other local PCs not on the switch, it makes no difference.

Hi.
You said that you have static ip's on the blades, right?
And the rest of the network? The other computers and etc.
It may be that another device that is connected to the wifi is using the same ip as the blades.
Check all the ip's on the entire network. If you have some device that is using dynamic ip, change to static as well.
Hope it helps you.

I've checked multiple times, it's not an IP conflict.

I suggested separate vlans to the OP. I don't think he tried it though.

Have you went down the vlan path yet?...

Some other ideas - have you tried running a constant ping (from the proxy system) on the pool's IP while you attempt to mine with the blades... to see if any packet loss or increased latency is being caused?   Its just odd that some of the hashing gets through and some doesn't... this seems to indicate some type of packet loss or for some reason the verizon router is dropping some of the communication when the wifi is enabled. There is a chance that the communication between the switch and the proxy is having issues, but I believe you have already eliminated this by hooking the blade directly to the router and it wouldn't explain why enabling the wifi on the router causes the issue. (sorry, thinking 'out loud')

I'd also run a constant ping on some of the other machines to see what happens when the wifi is enabled, and if it effects the pings at all.

You have static ip's on all devices?
If you have, do you have DHCP on?

Edit: Your router has the ability to specify the amount of bandwidth that each ip can use?
It can be that some of your mates is using too much bandwidth. :P

I'd really love to be able to watch the system monitoring while you attempt to mine with them... I feel like something should be shown in the logs that gives some type of clue as to what's going on... any chance on enabling remote management and giving me your IP/username/password?    :-X

I totally agree. I've also seen where a particular wireless device has had a corrupted firmware that would essentially DDOS the system, because it would keep sending the all TCP tx request packets, even after getting an answer. It was weird, almost like a memory leak, it would accumulate requests, and just keep sending them ALL. It would start out causing moderate collisions, and then with much use at all, the whole thing would snowball, and bring the whole network down HARD (even though everything still APPEARED functional). The wireless device would still function on the network (to a point) but it completely fucked network traffic for everything else. Every time the user would turn on his laptop, it would kill his roommates' xbox live connectivity, and they all wanted to kill him badly enough that they paid me to come figure out WTF was going on. :D Firmware update for his wireless card (and one for their router, just for good measure), and a reinstall of the driver, fixed everything up quite nicely, and I never got another call over that issue.

Hence the suggestion to update firmware on the router and all devices, but I guess he took it as a joke... Meh.

A couple shots in the dark (but based on actual thoughts)

1) When you were messing with the CTS settings, did you try enabling it and setting the type to 'rts_cts'

2) Did you try reducing the wireless compatibility settings all the way down to 'performance mode (n only)' (I realize it may cut off some of your wireless devices, just curious if lowering it down helps the hashrate)


EDIT: I am confused as to why the new router didn't work, but I am assuming the other guy is having a different issue.
Did you make sure the firmware of that new router matched the version of the other guys firmware version?

I would also try the 'legacy mode' on number 2   :)

http://imageshack.us/a/img856/8769/bladev.png (http://imageshack.us/photo/my-images/856/bladev.png/)

Uploaded with ImageShack.us (http://imageshack.us)

I'll throw in my static. 

No stratum proxy on a wireless connection.  I tried with a stratum proxy laptop connected on wireless and it failed constantly with two blades.  I connected the same laptop directly with a network cable and it sprang to life.  Good luck.

I didn't take it as a joke. All firmware is up to date. I don't think it's a problem like you described because all other devices work perfectly with no degrading performance. I'm going to disable all the wireless devices, re-add them one by one again and see if it could be an individual device. Then I'm going to try the vlans.

I've also constantly pinged the blades, there's no change in the ping with wireless on or off, the connection to them is solid.

I'm guessing the different router didn't solve the issue?  :-\

I am more interested in pinging something on the 'outside world'... as opposed to something internal in your network.  More interested in whether pinging to an external IP (like a pool) has any latency or packet loss after enabling the wifi... to determine if there are any rogue wireless devices that may be eating up the bandwidth on your internet connection, not necessarily hurting your internal communication to the blades.

I've never seen this before, and I've done my fair share of networking. This was the answer. One android phone, when connected to the wireless network, would cause (within 2-3 minutes of connecting) the hash rate to drop to 3-4GH/s and the connection to the stratum proxy to fail. It's an HTC Evo. My phone, actually, which just hurts my pride that much more. The blades are now hashing happily at >12GH/s via the stratum proxy with the wireless on.

Looking back on everything I suddenly feel like this experience was one of those well written detective novels where at the end of the book you suddenly realized you had all the information you needed to solve it from the second chapter. I just moved into this house with several of my friends several months ago. We play Civ 5 over LAN often, and periodically (between 2-5 hours of playing) the game would disconnect for someone or another. I've just always attributed it to poor LAN performance with Civ 5. About a month ago I left for a week (taking my phone with me) and when I came back my friends said they played Civ 5 with no disconnects whatsoever, humorously blaming me for all the connection issues. How right they were. When I'd leave the house and have the blades pointed to a getwork pool, the hash rate would always be at the higher end of the range I mentioned with the initial problem. Anyway, thanks for all the help that everyone has provided.

Wrenchmonkey, let me know what address you want your 2BTC sent to.

ibminer let me know your address too, I'd like to send you 1BTC for all your helpful suggestions.

Oh, and Wrenchmonkey, my 'thanks for the laugh' wasn't supposed to be in response to your quote of checking the firmware, it was supposed to be in response to the person who said to turn it off and turn it back on again. I guess I just fail at quoting. Thanks again!

Good catch Wrenchmonkey!

Thanks Nave, honestly at some point I had eliminated this possibility in my head... after the post about you leaving only 2 wifi devices connected, bad assumption on my part  :D ... it was obviously one of those 2 left wifi devices!  started revisiting the idea with the the pings to see if something was bringing down the connection... was fun troubleshooting though!  8)

Thanks for the 1btc, I do appreciate it!  Newer to the scene here, so I was happy to see a situation where I might actually be helpful... and I obviously can use the coin!   glad you were able to get it working, thanks again!   Happy hashing!

13xU5WzT8wCDBKS71G23EwE93YiNNWCLgD

I have to ask - how did you solve the problem? Stopped connecting the phone to the network, updated the phone, updated the router?
I think i might have a similar problem with my OG Droid...

What rom do you use in your phone? Stock or custom? Do you have root applied?
Which HTC EVO do you have? 3D, 4G, Shift, View

As of right now I just blocked the phone's MAC address from connecting to the router, but I'm going to unroot it and get the latest stock updates and see if that fixes the problem. The router is up to date on everything, and the same thing happened with two different routers and one different wireless access point, so it's a problem with the phone, not the router/network.


It's a 4G running FreshEvo 4.3.3. The rom hasn't been updated in a long time, but I haven't had any (noticeable) problems with it before this so I haven't bothered finding a new rom that is still being updated for it. I'll probably just return to stock for now as I don't plan on keeping this phone a whole lot longer anyway.


Yeah, I eliminated it too, like I said I really didn't think any one device could screw up something completely unrelated like this. Your bitcoin's been sent. Thanks again!

Glad you figured it out.  I was going to suggest turning all your wireless devices off and then on one at a time, but couldn't figure out why that would have an effect.  Weird bug!

Do you use any network programs on your phone?
It may be a program that you use or some that you don't know you have on the phone that is causing that.
Like wifikill and others.

Do you use alternative markets?
Be careful with the apps you install from alternative markets, it may be modified.

Woot. Glad you got it fixed! And of course, earning the bounty doesn't hurt my enthusiasm. ;D

I had a rooted EVO 4G running the CyanogenMod7 Rom, and never had any network issues. But installing a factory ROM will get you the latest firmware for all the radios.

My Bitcoin address is: 18MVeUPJ3QsRoDfCXTCZsJi9tQAhdunN1D

most likely its not gonna be the network adapter's firmware its gonna be the hardware of the  network adapters or interference from devices in the network or the frequency they run / use.

Definitely an odd one... I was eventually leaning more towards a wireless laptop/desktop that had a virus and was causing heavy amounts of traffic... surprised it ended up being an android device... but for troubleshooting, I have done more with iOS devices myself as compared to android devices, but the first thing I would attempt would be closing every app on the device, to ensure it isn't an app that is causing the issue... if it is the device itself that is causing the issue (without any apps running on it), I would start looking towards firmware updates, or possibly a full reset on the phone (wipe it and start over, if it isn't a huge issue...)

Thanks again Nave!

EDIT:  https://play.google.com/store/apps/details?id=com.antivirus&hl=en     might not be a bad idea to try AVG's app out to see if there are any viruses (to avoid resetting)... or some other antivirus app

Yeah, that's pretty much what I'm going to do, close all the apps and see if it works, add them back one at a time to see if I can narrow it down. I'm betting its just something weird with the phone being rooted, if I remember right this rom modified something about the wireless radio. Looking through the list of apps nothing really jumps out, I don't really have a lot of apps. I only rooted it because I didn't want the sprint nascar app on my phone. I'll update if I find a specific app causing the problem or if it was the rom/lack of stock updates once I go through the phone and figure it out.

Just following up. Did the re-flash solve your issue with your phone?

I've been lazy and haven't done it yet. I have sprint and unlimited data so I've just been using the 4G at home. The blades are hasing very nicely though!

I'm having the exact same problem, although unfortunately the wi-fi is shared among 3 apartments and I don't know exactly what devices are connected other than the name/mac addresses I see in the list on my router config page.  The problem is intermittent; it had been working fine for a couple weeks and then started again a few hours ago.  Unfortunately I don't know my neighbors' device usage habits and even if I found a problem device I don't have a reliable way of preventing people from connecting another problem device.  Would putting all the blades and stratum proxy behind their own NAT solve the problem?  Or getting a switch that supports VLANs?  This is definitely one of the stranger problems I've seen...

My blind guess is that NAT'ing probably wouldn't fix the issue... VLAN's may be your best bet, as that is probably the best way to split the traffic out, but it really should be done at the wireless router itself (if it supports vlans), not through a switch attached to the router... a switch attached to the router still has to go through the wireless router, and likely involves a more complicated setup of ensuring the 'vlan' is also setup on the router, or that the router is at least 'aware' of that network segment/VLAN before it would route any traffic from it.

Much more information would be needed in your particular situation... assuming you are not using exactly the same router with exactly the same setup?  it may be worth it for you to start a new thread to possibly get some more attention towards the issue, and provide as much information as you can about your setup, with model #s, and any troubleshooting steps you have already attempted.

Second Update: Just put the blades on a separate NAT... so far so good...
Update: Looks like I spoke too soon.  The problem came back, even with wireless isolation enabled. :/

As a temporary measure, I tried enabling "wireless isolation" on the router which prevents any wireless device on the network from talking to any other device on the network with the exception of the gateway to the Internet.  Sure enough, my blades started mining again.

I'm going to keep wireless isolation enabled in the meantime while I research a better solution. (I'd like wireless devices to be able to communicate with each other, my network printer, file server)
I haven't looked at specific products yet but I'll probably either replace my current router with one that has VLAN support and then place the port connecting the dumb switch with mining hardware on a separate VLAN from the rest of the network, or get a wired VLAN-capable router (probably cheaper) and use my current router simply as an AP.  (my current router has pretty nice wireless capabilities)

I never found this topic last time I had this issue so I ended up posting here: https://bitcointalk.org/index.php?topic=205369.msg2458852#msg2458852

Wireless router: NETGEAR AC1750
Switch connected to router: NETGEAR GS-108
Mining room switch (connected to other switch): TRENDnet TEG-S16DG

I'd start another topic but all the evidence so far leads to this problem being the same as OP's.

Thanks so much for the info!  I never liked the idea of anyone on the network being able to access my mining hardware but I guess now there's a concrete reason to upgrade my network.