Question

Weekly Reboot



Show first post

61 replies

Userlevel 3
Badge +4

I make it a habit of checking my modem’s uptime on a daily basis.  I’ve been waiting to get past that 7 day milestone now that I have the cooling fans installed.  Yesterday morning I was over 5 days.  Throughout the days we had some strange internet problems around the house.  Sites would load from one device but not another.  Then eventually it would work again.  I would check to make sure pings were still working and they were.  Speed tests came back normal.  I attributed it to the worldwide issues from the day before.  Last night I checked on the device again and, much to my surprise, the uptime was only about 16 hours.  Assuming this was correct, it had rebooted before I was even up for the day.  This would also mean the uptime counter I saw earlier in the day was not accurate.  I think it’s cached in the browser a bit and if you don’t click refresh it could show you the wrong time.  So as best as I can tell, the modem restarted yesterday morning.  It has not done that on me before.  But with the weird issues experienced throughout the day, my guess is that it was maybe running low on memory.  We have 20 devices online and all that NAT probably wears it down over time.  So I decided to make the change back to my previous wifi router and only use the modem as a gateway.  I also decided to move it back to the basement.  The tower it connects to is close enough that I still get decent signal from the basement with the added bonus that it’s cooler down there than the rest of the house.  I’ve also determined that even though my secondary signal uses a 5G band, it’s not actually being used.  So it’s basically 4G home internet which I am happy with as long as it’s stable.

Userlevel 3
Badge +4

I hit a new milestone today: 14+ day uptime!  My device is back in the cooler basement but I never got this far without dedicated cooling fans.  I have the FCC speed test app on my phone and it runs daily speed tests in the background.  It’s claiming an average download speed of 99.5 Mbps over the last week with an average 44.5 Mbps upload so I’m happy.  The only complaint is that a couple days ago my public IP changed overnight.  Any network connection I had open from my computer was dropped but as far as I can tell it was not a device reboot.  If I had to guess I would say the DHCP lease expired and it could not renew my IP for some reason.  I’m not 100% sure but I will keep an eye out and see if that happens on a regular basis.

I read that you mentioned heat as a issue and I would be inclined to agree that this might be an issue. I used to have an old phone which I ran a tethering Hotspot to my computer to share the connection with my other devices through Ethernet. When the phone got hot, due to it charging and processing internet data at the same time, it would lose all connectivity to the internet. 

I eventually got a new phone that ran much cooler and generally I could have a uptime as long as I wanted to, or at least until the tethering app crashed, but that was fixed by rerunning the app and reconnecting. 

Now I had Straight Talk which runs off of the same T-Mobile/Version network so I had a general idea on the quality of internet I would get, even though its generally 2 to 3 bars of 4G only. However, so far when the service works I get vastly superior speeds to what tethering my phone was capable of. The downside is that I experience disconnects that are fixed by rebooting the device, similar to your issue, or just waiting it out. 

The secondary signal does sometimes disconnect for me, but generally I either still maintain my service connection regardless, or when it goes offline I still see that my secondary signal is still connected.

Mine generally runs 8-32 hours of uptime before a restart, or multiple restarts, while running only around 6 Wi-Fi devices and 1 or 2 Ethernet connections

Since you have issues with a 4-5 bar connection I think its safe to say that my issues are generally not brought on by occasional poor connection.

When I place my hand over the vent I can tell its pushing out hot air weakly, and as you’ve said in your latest post those cooling fans seemed to have helped. My device is in a small room with my PC and Consoles so it does get hot in here depending on what I’m running since our house naturally hovers around 75 degrees. 

Also as a suggestion you can try with the Wi-Fi when it comes to connecting multiple devices. Try separating the 2.4Ghz and 5Ghz Wi-Fi bands in the settings. When I first got my device they were sort of blended together and separating them would help with connectivity since 2.4Ghz and 5Ghz don’t interfere with each other due to being different frequencies. I believe all you need to do to separate them is rename the SSID of the 5G just to say 5G at the end and make sure the password is the same as the 2.4Ghz, unless you want a different password of course. 

Side Note: I think the maximum devices for each Wi-Fi band is 10, I don’t remember where I read it, but I think it was in a manual. Though I would think that 20 devices running off the same frequency would cause interruptions.  

Userlevel 3
Badge +4

Also as a suggestion you can try with the Wi-Fi when it comes to connecting multiple devices. Try separating the 2.4Ghz and 5Ghz Wi-Fi bands in the settings. When I first got my device they were sort of blended together and separating them would help with connectivity since 2.4Ghz and 5Ghz don’t interfere with each other due to being different frequencies. I believe all you need to do to separate them is rename the SSID of the 5G just to say 5G at the end and make sure the password is the same as the 2.4Ghz, unless you want a different password of course. 

Side Note: I think the maximum devices for each Wi-Fi band is 10, I don’t remember where I read it, but I think it was in a manual. Though I would think that 20 devices running off the same frequency would cause interruptions.  

I used to agree with separating the 2.4Ghz and 5Ghz wi-fi networks (I’ve been in the networking business for decades) but if you have good hardware that doesn’t need to be the case any more.  I bought a TP-Link Archer C6 router last year before I got T-Mobile and it’s been great.  I use the same SSID for both networks with no problems.  I got this router because my previous provider’s all-in-one device did not do well on the wi-fi side.  I found the same with T-Mobile so I’m not even using the wi-fi on the Nokia.  I’m using a single ethernet connection between the Nokia and the TP-Link and have turned off all wi-fi SSIDs on the Nokia to make sure it doesn’t get in the way.  With all that I’m looking at a stable 4G hotspot with a 15+ day uptime.

Userlevel 3
Badge +4

I had to reboot my modem last night after staying up for a good 18 days.  I work from home and it’s not unusual for me to download or upload files that are over 1 GB in size.  Usually that’s not a problem.  Last night, however, I found myself downloading one of those large files while also uploading a large file.  I would say this does not usually happen.  In the middle of the transfer everything completely stopped working and I had to reboot.  Once I was back online I was able to continue going with no problems.  I wonder if all the bi-directional traffic going on at the same time was too much for it to handle for too long of a period.

I haven’t gotten around to troubleshooting issues much since I’m waiting for a USB fan to come in the mail to try and resolve the heat issue. Last night it was uncomfortably warm and was either offline or online just enough to struggle to load webpages even with multiple restarts and power downs.

I will say though that when it is working I do notice during session where I upload and download at the same time the download’s performance is unaffected, yet the upload performance takes a massive hit. I would also get unusually low Upload speeds on tests early on when I had the service. Like Kbit/s range while the download was anywhere between 50Mbit-100BMbit. I figured it was because my signal hovers around 2-3 bars, but I just ran it today and got this score.

 

Express Test on Testmy.net

Even tested large file size of 100MB and even though it took a minute it still said it was around 9.0 Mbps

 

latency is low, but does generally get worse if I’m running a video in the background, though its not enough to cause issues unless you’re transferring files to London or Frankfurt. 

----------------------------------------------------------------------------------------------------------------------------------------

Okay so while that same video was still running in the background I tried a speed test. At first it scored the same, just slightly lower 8.5Mbps. But check out what happened after I let the video run for a bit. 

 

The video was only at 480p by the way. I don’t know why, but regardless it is effecting the upload severely. I tested 3 times and got similar results. I test 3 times after the fact and still got these slow speeds and even lost connection to some of my websites while going through my tabs and refreshing. Didn’t need a restart though.

 

Download however, during my testing with the video.

 Is completely unaffected. And no everyone else in the house isn’t doing anything extreme. Just web browsing while I do these tests and the router isn’t even hot yet. I tested after the fact too and its still just as fast. 

Userlevel 3
Badge +4

I had things freeze up on my last night.  It was actually quite strange.  I suddenly lost all connections to a particular set of servers I was working with.  They are all in the same datacenter so at first I thought it was host issue.  I couldn’t hit the hosting provider’s website either.  I was still able to access Gmail and some other Google sites.  But using Google’s built in speed test tool failed.  I could still ping their DNS server though (8.8.8.8).  Just in case it was a T-Mobile issue I dropped my phone off the wi-fi and everything worked fine.  If I didn’t know better, I would say the issues I saw were similar to the ones I would experience when I tried using the devices’ built-in wi-fi.  So even though I have a dedicated wi-fi router there must be something on the device that is still leaking memory over time and eventually starts failing.  I did get to 21 days of uptime though which is a new record.  With this as a possibility when I leave the device up for too long and the fact that I can’t keep the same public IP for more than 2 weeks I will probably start doing a planned reboot every time my public IP changes which means, in the best of times, a reboot every 2 weeks.

 

 

Userlevel 2
Badge +1

With the information you have posted so far it is hard to say for sure as you need to look at the statistics a little closer for the signals. Sorry, I am not trying to be a pain or jerk just trying to help. I have been fighting similar issues as well. I know my GW is on software version 1.2003.03.0178 and that the unit does pick up two channels, the LTE (B2) and 5G n71 channel. You report seeing 5 bars on the can which is 1-2 more than I see so that is positive. I know when my GW reports 3 bars the tower signals are not being received as well. Reporting via the software connection vs the top of the GW seem to be a little out of sync for me. T-Mobile states it only supports up to 64 devices in the configuration of the GW and more devices vs fewer might lead to some throttling. I am not sure. I did have more unstable behavior with more clients. Not sure if it was related or not. For me the wireless signals for local WIFI look great but then random out of the blue no internet reported. Well, it relates to the tower uplinks. The primary signal tends to be the LTE signal and the secondary signal is the 5G band or that is what I have seen every time with mine. If the n71 is not reported it gets slow and of course more latency with the LTE signal. I reviewed an article that seemed to suggest T-Mobile uses a blending of the LTE and 5G signals so attempting to go with just 5G is probably not going to happen. I see no way to influence the signal handling of the connections to the tower. Look at the RSRP, RSRQ and SNR values. If you have good or excellent RSRQ signal quality that will be positive. Of course signal strength reported as -100 dBm or better say > -90 dBm is excellent but if the signal quality is < -13 dB then that is fair to poor and will impact the signals. These are values to understand and know. I have spent considerable time on this and I know I have good signal strength RSRP and good to excellent RSRQ signal quality and the SNR signal to noise ratio is excellent but I still see random signal drops which is frustrating. A T-Mobile support engineer could tell you if they are doing work in the area on the towers and given the aggressive push T-Mobile is making on expansion could be a factor. Regarding airflow, well I am in TN and I put my can outside in 88 degree heat to improve the signal to reduce possible interference and it is running fine and holding signal for hours now where it was dropping one or both channels to the tower rather frequently. I have the unit in the shade but the 88 degree temperature does not seem to phase it.

Userlevel 2
Badge +1

I have had the T-MO gateway running here since early in January so i have seen several updates in the software now. I don’t know for sure the version it was running just prior to 1.2003.03.0178 but until the past few weeks I had not experienced many issues with connections to the towers at all. In fact it was very stable and seldom needed to have any attention. I have my suspicions about the 1.2003.03.0178 software version because the random tower disconnects are more recent. Before  rebooting I recommend recording the tower uplink information to record the RSRP, RSRQ and SNR values over time and know what channels are used. I see a few dBm difference here and there but the main factor I see as impact is the loss of the n71 channel/band on the secondary signal. You really want to see both the LTE and 5G signals standing. You might find, as I did, when only the LTE signal on B2 is up it becomes unreliable. Recording the information about the uplinks to the tower and having a historical record would probably be helpful. The more you know the better the profile of operation and behaviors. 

Userlevel 2
Badge +1

FYI - You probably know this but well just in case. Supported T-Mobile bands in the US.

  • 5G Bands: n2, n25, n41, n66, n71
  • 4G LTE Bands: B2, B4, B5, B12, B25, B26, B41, B46, B48, B66, B71

There is an interesting article at waveform.com you would find helpful. (Excellent information)

  1. Hotspot Guides
  2. T-Mobile High Speed 5G Internet Gateway External Antenna Guide

No I have no connection to them. I found this while picking this problem apart. This is some of the best information I found when researching the T-Mobile Nokia gateway.

Userlevel 2
Badge +1

Just a statement an excellent engineer told me long ago. “I don’t want to know what you think. I want to know what you know.” I took that to heart 20 year ago when I ended up taking his position on the L3 hostile team. I respect him a great deal and that helped me be successful for the past 20 years as a network escalation engineer. It is obvious to me you have been working hard at picking this apart. You might be onto something with the software and a memory leak or some table corruption. I had very few problems with uplinks to the towers between early January and into late June. I don’t know when the software version 1.2003.03.0178 was pushed out so I don’t have enough data to have a full picture on this but it does seem suspicious to me.

Userlevel 2
Badge +1

So you are running the TP-Link Archer C6 I get that but you are aware the Nokia is 802.11ax capable. I thought of getting another router to link to the Nokia just to have better visibility and more control of the clients as the Nokia is a bit limited. Do you receive the 5G band or not? you talk of the device as a LTE router but it should have both LTE and 5G signal links. I found when mine drops only to LTE it tends to be much less reliable than when the n71 channel is also up. I would for sure monitor the signals that link up to the tower. 

Userlevel 3
Badge +4

I ran into a new issue but I think I’ve seen a few other posts touch on it.  Yesterday I noticed my connection seemed slower than usual.  It continued today and I finally considered doing a speed test to confirm.  It was not dead in the water like my usual overheating scenario so I did not immediately assume it was a modem issue.  My speed test came back with around 10 Mbps download.  Yikes!  I have the FCC speed test app on my phone and it does background checks once a day.  It shows that my speed dropped from 72 Mbps on Sunday to 5 Mbps on Monday and 2 Mbps today.  So I checked the modem and noticed that my primary signal was on B2 and my secondary signal was not connected.  A quick reboot brought me back to B66 / n41 and my usual speeds.  It would be nice if the modem would keep an eye out for better signals and switch back when possible.  I understand falling down from time to time if something is unavailable but why not pop back up too?  Here are some before and after screenshots.

 

Userlevel 3
Badge +4

After posting my screenshots I realized that I actually don’t usually get n41 for my secondary signal.  I am usually on n71.  I can confirm 200 Mbps downloads now so maybe something has changed on the tower in town and I’m getting better reception now.  Woot!

Userlevel 2
Badge +1

So you are on the n41 5G channel and your unit reports -108 dBm so that is in the Fair signal strength range. The -11 dBm puts the signal quality in the excellent range and the signal to noise ration is 13 dBm so that is still in the good range. I forget the range of the n41 band. I would have to look that up myself. I can't keep these things in my head. I do recall in my research seeing estimated range values for various 5G channels. I had some success with the rotation of my router making the 5G n71 come up 3-4 dBm and the 4G LTE went down but the end result was the LED panel on top of the router reports 4 bars vs 3 when I spin the can to have the 5G so it is most exposed in the direction of the tower. That is interesting. We appear to have the same hardware revision 3TG00739AAAA. Good or bad I dont know but we for sure use different primary and secondary channels.

Userlevel 2
Badge +1

OK so interesting. My understanding, unless my memory fails me, is that n71 has more reach than the n41 band.

Userlevel 3
Badge +4

Yes.  Lower frequencies go further but are slower.

https://www.t-mobile.com/support/coverage/t-mobile-network

Userlevel 2
Badge +1

I use a Linux client and run concurrent PINGs to google, guad9 and cloud flare DNS servers to compare latency. Run them and then record the values you have for the primary and secondary channels. I would bet you will see a nice change on the n71 channel with the 5G. Your download speed test is pretty good. Mine currently will pull 127 MBs down and 35 Mbs up with 54 ms latency to a server in KY. That is pretty consistent for that server.

Userlevel 2
Badge +1

Right but what I found on waveform.com, is that signal quality plays a big role in throughput and I believe this as if you have lots of bandwidth but errors you will have retransmits due to the basic operation of TCP/IP. Clean is always better of course. I would bet they did an upgrade on the tower your device is associated with.

Userlevel 3
Badge +4

I’ve increased the frequency that my scheduled speed tests are run so I will have a better idea of if / when things change.

Userlevel 2
Badge +1

From what I have seen with my T-Mobile router is that it will do ~150-170 Mbs down in periods when the traffic through the tower is not as heavy. The spin of my router to expose the surface to the left of the back side of the router, i.e. where the LAN ports etc… are allowed the 5G to improve and the LTE went down a bit. I think the radial radio antenna design in this router maybe prevents getting the best signal wash for both LTE and 5G signals.

Userlevel 2
Badge +1

I was not really running scheduled periodic speed tests. Well, given mine was dropping both channels it would have been clear when the disruptions were but running the concurrent PINGs was helpful as the lag and flakey response of the HTML reporting by the router to the client did not give me a real accurate picture when it actually was down on the uplinks. It might be a cashing and refresh issue in the software. I can see the real time clock on the one view running but that I am not sure matters much.

Userlevel 2
Badge +1

So your router receives the n71 channel how do your RSRP, RSRQ and SNR values for the 5G now? Probably the RSRP has come up a few dBM.

Userlevel 2
Badge +1

From helping my neighbor out with his router I discovered some interesting information about the design of the can and made this diagram. I believe the antenna labeled 5G-4 is the one that the n71 channel is received upon. The tower here is due North so I spin the 5G-4 to have the best exposure and that helps in this location.

 

Userlevel 3
Badge +4

I just checked cellmapper.net and it looks like a couple n41 cells were added to our tower over the last couple months.  I’m guessing that’s what I’m attached to now.  I will probably leave the modem alone for now but if I fall back to n71 I may try some different rotation angles to see which one works best.  Your diagram will help with that.  Thanks!

Reply