TP-LINK WiFi nano router with mobile hotspot

I have a Pi-Top with R-Pi3 and TP-LINK router, model TL-WR702N v1.2, inside, so that when on the move I can access the internet via my Netgear mobile hotspot.
I have succeeded in getting it to access the internet via WiFi to the home router (usual wired connection to telephone line), but failed when using the mobile hotspot in place of the home router.

I can ping external websites from the Pi-Top’s CLI or a task window successfully, but cannot get an external web page using its IP address in a browser.

I see from earlier posts that some folk have succeeded, for example, Jon Abbott, Steffen Huber and Dave Higton. The TP-LINK seems to have the ‘right’ build (130314) according to the web site referred to by Dave Higton, and I have followed the instructions for installation given there including using Client mode.

I have changed a lot of the IP addresses and one DHCP range for consistency with my home LAN, for example I use a subnet mask of 255.255.0.0.

My ASUS pad running Android which has WiFi built in, but not access to the cellular network, accesses the internet well via the mobile hotspot both before and after the changes. It’s presumably getting an IP address from the DHCP server on the mobile hotspot, whereas I have a static address for the Pi-Top.

I would really like to get this working so that I can use RISC OS on the move, and this is important for our OS.

A common problem with DHCP on RISCOS, is that it fails to set the DNS server and/or gateway. I suspect this is the issue in your case, type show Inet$* and see if the correct values are in Inet$Resolvers, Inet$IPaddr and Inet$Gateway

As far as I can tell, DHCPExecute doesn’t handle the DHCP offer it gets from certain devices correctly.

Setting a gateway address upsets DHCP!
The gateway entry must be left blank when using DHCP.

Setting a gateway address upsets DHCP!
The gateway entry must be left blank when using DHCP.

A couple of things:

1. I believe that Jon was pointing out that the implementation of the DHCP server function in some routers is not quite standard and that the RO DHCP client implementation is a bit sensitive to such things and sometimes fails to recognise the DHCP provided gateway.
2. Every IP related network setting (IP, mask, gateway, name servers) should be null for DHCP to operate correctly.

I’d like to help, but your settings and mine are very different. I use fixed IP addressing based on the host name and the hosts file. I have an Obey file that I run when I’m setting the machine up with the nanorouter and hotspot – it changes the few things that are different, and doesn’t require the machine to be rebooted with different settings. I don’t use DHCP.

Looking back at the file, I see that most of the differences between home and hotspot use have disappeared (as a result of changes I made to the home LAN); the remaining one is that the gateway is 192.168.16.1 at home, and 192.168.16.254 with the nanorouter and hotspot.  My netmask is 255.255.255.0 in both cases. I believe that the gateway address is determined by the mobile phone. For me it’s just lucky that the IP address range is the same in both cases.

The most surprising thing to me about your setup is that you use a netmask of 255.255.0.0 – it isn’t necessarily wrong, but it is highly unusual, so it needs to be questioned.

@Jon
Show Inet$* gave the correct values for the resolvers and IP address of the computer: For a wired computer and for the pi-top via its TP-LINK WiFi adaptor. In both cases the Inet$IsGateway value was blank. Both have static addresses and for both computers the setting “Act as an IP router” is unticked. The LAN router uses the telephone line, not a mobile hotspot.
@Dave H.
I have changed the netmasks on router, TP-LINK nano gadget and computers to 255.255.255.0. (The reason for having 255.255.0.0 was that I had some gadgets/routers/computers with addresses 192.168.0.X and others with 192.168.1.Y and I got fed up with constantly having to change addresses from one type to the other to make things work; but I am now changing every machine to 192.168.0.X)

The problem remains getting the pi-top to work with the mobile hotspot. Now that the mobile hotspot and wired router have the same IP address, if I give a computer a static address outside the DHCP range of either, then I should be able to keep the settings in the Network section of Configure and only need to change settings on the TP-LINK nano gadget when changing from wired router to the mobile hotspot or vice versa?

I am assuming that the DHCP address ranges do not need to be the same on each router’hotspot or TP-LINK nano gadget, especially as DHCP is not being used.

I am assuming that the DHCP address ranges do not need to be the same on each router’hotspot or TP-LINK nano gadget, especially as DHCP is not being used.

As long as you don’t have two DHCP servers(sources) on the network at once.

That said a few years ago the MS recommendation for dual/resilient DHCP installs was a 70/30 scope split with the same gateway details etc but one server only handed out the bottom 30% of IP’s and the other did the upper 70%.

In your case with manually assigned and DHCP you’d just assign say 50% of the IP’s for the DHCP pool and then manually assign¹ the others to problem devices

NB. With two servers for the same subnet the first to offer (DORA) gives the device an IP.

¹ I prefer the label manually assigned as a description as a static address could be from a reservation in DHCP or manually assigned.

I can ping external websites from the Pi-Top’s CLI or a task window successfully, but cannot get an external web page using its IP address in a browser.

I really can’t understand how that can happen unless there is a firewall rule somewhere that prevents HTTP access in some way. Any ideas from anyone else?

I don’t see how it can be caused by a DHCP problem. AFAIUI ping requires the gateway to be configured and working, just as much as does HTTP or anything else.

…only need to change settings on the TP-LINK nano gadget when changing from wired router to the mobile hotspot or vice versa?

I don’t understand what you’re trying to do. If you are on a wired LAN, you ought not to have the nano router active at all; if you do, you will have an IP address clash for the gateway unless you change the nano router’s settings in some way so as to disable it. The gateway address (again AIUI) is in the mobile hotspot, not in the router; you may not be able to change it.

@Dave H.
To be clear: I can access the internet successfully using the ‘home’ router connected to the telephone line using WiFi between that and the Pi-top wth its TP_LINK nano gadget. I have also acheived the same with a different ‘home’ router. But I have failed when using a mobile hotspot or ‘smartphone’ in place of the ‘home’ router. It is the latter that I am now trying to get to work.
If, in the above, I replace the Pi-top with my pad running Android, the pad can access the internet.

I have a static address for the Pi-top, so not DHCP setting, in the RISC OS Configure Network settings. The TP-LINK gadget is set to Client mode with DHCP disabled. The IP addresses of the TP-LINK gadget and Pi-top are different, different from that of any router or the mobile hotspot and not in the range of any DHCP pool on routers or mobile hotspot. All, except one, have the same netmask, 255.255.255.0, and all IP addresses are of form 192.168.0.X. The one exception is the ‘smart’ phone that has IP addresses of form 192.168.43.X. The smart phone has “network prefix length” 24 which, I assume, means netmask 255.255.255.0 as 2 to power 24 is 256 cubed.

Both the mobile hotspot (communicates with internet over the cellular network) and the smartphone acknowledge the presence of the Pi-top, or at least the TP-LINK gadget, showing that 1 computer is connected with the smartphone giving the MAC address of the TP-LINK gadget. Pinging an external website has sometimes succeeded and other times resulted in one, almost complete, line of packet return information (i.e. success) followed by a total crash: “Internal error, no tact for strap handlers …….abort on fear of death …” :-((
Even when an IP of a URL can be successfully pinged (e.g. RISCOS Open), I cannot get the corresponding web page in a browser ending with message “Timeout was reached”.

My feeling is that the mobile hotspot and smartphone are only in DHCP mode and so unable to deal properly with the TP-LINK gadget on static address. If so, I need to alter the settings on mobile hotspot and smartphone to deal with static addresses. My problem is that I am not very good at experimenting with devices due to the fear of screwing everything up. I know I can do factory reset and lose all my settings. :-((

The smart phone has “network prefix length” 24 which, I assume, means netmask 255.255.255.0 as 2 to power 24 is 256 cubed.

Yup. /24 masks the first 24 bits of the 32 bit (IPv4) IP address.

Even when an IP of a URL can be successfully pinged (e.g. RISCOS Open), I cannot get the corresponding web page in a browser ending with message “Timeout was reached”.

Ping to the known IP (91.203.57.172) or www.riscosopen.org?
If you can do the former but not the latter then you probably don’t have the right DNS settings.

As Dave pointed out the ability to ping anything out there shows the IP and gateway setup is correct. If you can ping local but nothing beyond your gateway then the subnet mask is the issue but in this instance that seems OK so you’re on to the name resolve element which starts with a query to a listed DNS instance.

Note that some ISP’s restrict access to their DNS servers to their subscribers so if you have that problem and want to switch between provided routes and always use the same DNS then use 8.8.8.8 or 8.8.4.4 as one entry in your DNS list.

What are your “Name server” settings (primary, secondary and tertiary) when you use the wired lan, and when you use the nano router?

I really can’t understand how that can happen

I’ve sometimes seen Ping working but browsing using ip address not, if the netmasks is inconsistent..

I’ve sometimes seen Ping working but browsing using ip address not, if the netmasks is inconsistent..

That would only be possible if the mask of the device was defining a smaller subnet that formed part of the larger subnet defined by the gateway router.
I don’t think it would work completely correctly though.
If you had two different /24 definitions (192.168.1.0/24 and 192.168.2.0/24 devises in them could be in the same switch and not communicate.

In general when people come to me with a “it won’t talk to things in another VLAN” type of query then the first suggestion is that the subnet mask needs looking at along with the gateway.
I also tend to point out that if they hadn’t been playing with manually defined settings instead of utilising DHCP it probably wouldn’t have happened.

Just occasionally though the supplied devices don’t work the way they should and manual intervention is required. It’s the exception not the norm.

I have finally got both the TP-Link Nano gadget (acting as a WiFi adaptor, i.e. client mode) and the Netgear WNCE2001 adaptor working with my mobile hotspot (Netgear Aircard-F878). (See my post of 11/12/16 above.)
One thing I needed to change was the DNS addresses in the “Host names” section of the Network configuration to just the gateway IP address of the hotspot (as primary name server) rather than the IP addresses (as primary and secondary name servers) of my ISP.
However, although this worked on one computer, it failed to work on a second computer with the same settings apart from the obvious chsnges for the LAN IP address of the computer and its name. Why?