hAP ac lite slow ethernet

Very low priority question but I use a hAP ac lite in the lab for experiments and learning. Happened to do a speed test from a laptop hard wired into the router and was surprised at the speed: struggling to get above 60Mbps. The router is connected upstream to a gigabit unmanaged switch. Both the router and laptop ethernet interfaces are reporting 100Mbps full duplex. Plug the laptop directly into the upstream router and Ookla speedtest gets the expected ~300Mbps of my internet connection.

I’ve just reset the router without a default configuration and set-up it up as a simple switch (add bridge, add all ethernet ports to bridge). Same speed so it’s not the firewall overheard that you get when using it as a router.

I know the hAP ac lite doesn’t have a powerful CPU (MIPSBE single-core 650MHz) and I didn’t expect 100Mbps but certainly expected above 90MBps.

Perplexed of Manchester…

PS. Also got a hAP lite and same speeds when configured as switch…
PPS. Found some very old threads on here about ethernet → ethernet speed problems on the hAP ac range. As I said, it’s just lab kit so not that worried, just curious. I guess as these tend to be used on FTTC connections which max out at around 60Mbps, it’s not a massive issue.

This is the switch config - doesn’t get much simpler than this:

/interface bridge
add name=bridge
/interface wireless
set [ find default-name=wlan1 ] ssid=MikroTik
set [ find default-name=wlan2 ] ssid=MikroTik
/interface wireless security-profiles
set [ find default=yes ] supplicant-identity=MikroTik
/ip hotspot profile
set [ find default=yes ] html-directory=hotspot
/interface bridge port
add bridge=bridge interface=ether1
add bridge=bridge interface=ether2
add bridge=bridge interface=ether3
add bridge=bridge interface=ether4
add bridge=bridge interface=ether5
/ip dhcp-client
add disabled=no interface=bridge
/system clock
set time-zone-name=Europe/London
/system identity
set name=APS001

ROS 7 or 6?

ROS 6. I do recall upgrading one of these to ROS 7 last year - they are experimental devices for me after all. I spotted that it ran visibly slower, e.g. doing a /export. I rolled them back to ROS 6 as this is what most of my clients are still running with a note to re-visit again.

What model is the client wireless adapter and what is the speed at which it connected or you only tested wired?

I tested wired and wireless. The Wi-Fi 5 speed was limited to about the same speed (60Mbps). From memory, the lites only have only chain on Wi-Fi 5 so that’s got a theoretical speed of ~450Mbps (?) but even that should translate on to nearly 100Mbps. The fact the wireless speed and Wi-Fi 5 speeds have the same maximum suggests it’s either a limitation of the hAP lite, bug or configuration problem.

Both tests will be downloading through ether1 either in router or switch mode. As it occurred on both a hAP ac lite and hAP lite suggests it’s not a faulty device. I’ll connect my desktop computer and Raspberry Pi later.

I’ll upgrade them to ROS 7 as a test as well.

Do you have fast track enabled?
I had hap ac lite and it works with 100mbps uplink with full wire speed.

Do you have fast track enabled?

I do yes in router mode. Interesting question here, when you have it configured as a switch without a firewall, is fast track even a thing?

I had hap ac lite and it works with 100mbps uplink with full wire speed.

Thanks, that confirms that the hAP lite can get near 100Mbps through the ethernet ports and that it’s not an inherent design flaw. I’ll start looking elsewhere like connecting it directly to my Virgin Media modem. At the moment the connection is:

Virgin Media (modem) - hAP AX2 (main router) - unmanaged gigabit switch - hAP lite - laptop

Divide and conquer…

No, fasttrack is firewall thing (specifically: filter part with connection tracking working; raw doesn’t relate to fasttrack).

Take a look at this https://wiki.mikrotik.com/wiki/Manual:IP/Fasttrack

You have something wrong with your network, maybe the laptop too:

Try to force 100 Mbps FD in NTB and connect directly to the switch instead of MikroTik. Most likely you will already have 60 Mbps here.

You have something wrong with your network, maybe the laptop too:

Think it’s the Lenovo laptop with Realtek network chipset. Plug laptop directly into upstream 1GBps switch - no problem. Replace the hAP lite with an old TP-Link 100Mbps unmanaged switch I’ve got lying around - same maximum of 60Mbps. Or 100Mbps devices connecting upstream to the 1Gbps unmanaged router. Either way, it’s not a problem with the hAP lite as such. Dim memory of problems mixing 100Mbps and 1GBps…

For the record this scenario is the problem:

Virgin Media modem > hAP ax2 ↔ unmanaged TP-Link 1GBps switch ↔ hAP lite ↔ laptop

With that (cheap) unmanaged TP-Link switch in the way, the hAP lite ethernet is throttled to ~60Mbps. Remove it and all is well (93Mbps):

Virgin Media modem > hAP ax2 ↔ hAP lite ↔ laptop

It’s also not a specific problem with hAP lite connected to the unmanaged switch. The same throttling occurs if one replaces the hAP lite with another TP-Link 100Mbps unmanaged switch. It’s not the laptop either as same occurs with Raspberry Pi 4.

Later…

I’ve replaced the unmanaged TP-Link 1GBps switch with another equally cheap (but new) unmanaged TP-Link 1GBps switch and same problem. Of course, both TP-Links could have the same design problem! Don’t have any other 1Gbps switches around. Will borrow a client’s Netgear one as a test.

Am I reading correctly that an unmanaged (bad) TP-LINK (bad, bad) switch with 1 GB ports has the same throughput of another (presumably older) unmanaged (bad) TP-LINK (bad, bad) switch with 100 Mb ports? :open_mouth:

Can you post the exact models involved?

Yup you got it - I’m not going to post the models as I’ll be embarrassed :smiley: Okay, new one is a LS1005G… interestingly I’ve just tried this:

VM modem - hAP ax ↔ new TP-Link unmanaged 1Gbps switch ↔ old TP-Link unmanaged 1Gbps switch ↔ laptop

Laptop is only getting a 100Mbps connected… will try with some real kit :wink:

TP-LINK switch toubleshooting guide:
https://www.tp-link.com/it/support/faq/737/
essentially they revolve around quirks in auto-negotiation of the connection speed.

And here it is an indirect confirmation:
https://forums.overclockers.co.uk/threads/local-network-transfer-killing-internet-speed.18976040/

If the given explanation about “head of line blocking” (towards the end of the thread) is correct the PC is trying to “pull” more data than the 100 MB port on the other side can deliver and this makes the switch “confused”.

Whilst I did say this was low-priority, there is a slightly serious point here. If a client site is on a budget or hasn’t had the best advise in the past setting up their network, they could have anything upstream. They might ask you to install a new access point for as little money as possible and speed isn’t that important. So you say they can have this hAP ac lite that you’ve got lying around for £10 on the assumption that as their internet connection is only 150Mbps, 100Mbps would be fine. But then you’re perplexed why it’s only getting 60Mbps because of this fault upstream.

Well this one has come back to bite me. Setting up an event and put hAP AC lite in for a help desk with laptop. Hard wired back to Netgear gigabit switch. Laptop hard wired into hAP. Speed struggling to get above 30Mbps. Plug uplink cable direct into laptop and full 200mbps of the Starlink.

Will investigate if I have time but that comment above about it struggling with 100mbit and gigabit mixed sounds true.