Asuswrt-Merlin - custom build of the Asus RT-N66U firmware

[DaveMishSr]

I actually didn't clear the nvram until I made sure that my 2.4GHz. wireless configuration page access was fixed by the Asus 178 firmware. After I knew that the problem was taken care of, I did clear the nvram before configuring the 178 firmware because I was going between a 64KB and 32KB nvram firmwares.

It didn't occur to me that I should clear the nvram for 178.15, since I had only turned the power off and back on. On the other hand, I have fixed problems in the past by reflashing firmware. Next time something like this happens, I'll try clearing the nvram before reflashing (after power cycling, most likely *smile*). Nothing to lose by doing that at that point, and would have gained a bunch of time if it had fixed the GUI breakage.

As a precaution I *always* clear the nvram before and after flashing. That way there should be no *bugs* to interfere with the flashover and the configuration should also be clean as it is being set up in a pristine, read factory, environment. Just sayin'.

 

[RogerSC]

As a precaution I *always* clear the nvram before and after flashing. That way there should be no *bugs* to interfere with the flashover and the configuration should also be clean as it is being set up in a pristine, read factory, environment. Just sayin'.

I don't disagree...I have found that clearing the nvram after an upgrade is enough. It doesn't mean that there aren't cases where it isn't, but I haven't run into any of those yet *smile*.

Thanks.

 

[NeutralGold]

Hey I just noticed a small problem with my client list on the network map page. As soon as the 10th client connects the client list hangs and won't display anything other than "Searching, Please wait.". The client count also shows 0 instead of 10. I did read a few early pages in this thread about hangs with the client list.

Just figured I'd ask instead of trying to sift through all of that.

I figured out the problem finally. I was playing around with ipv6 a day or 2 before this problem happened, but I also disabled ipv6 right after I was done as well. So it didn't occur to me that it could be the cause.

I've had about 90% memory consumption on my router for last couple of days as well. That was a little hard to spot for me since I keep transmission running and when someones downloading a torrent, memory usage is usually really high anyways. I didn't remember to start transmission after my last reboot and 150MB of ram was in use for no apparent reason. I checked "top" and didn't see anything running with high memory usage.

I was browsing around my router interface and noticed I left "Enable Router Advertisement" under ipv6 enabled. This caused the issue with my client list number to stop working after the 10th connection and also the very high memory usage.

Just thought I'd follow up with the cause in case this is a bug. I don't know enough about how ipv6 works but I just assumed since ipv6 was disabled that it negated the 2nd option of "Enable Router Advertisement".

It's only been a couple of hours since I spotted this, but so far I've not seen any raise in memory usage and it's now showing the 10th client and client list as it should.

Thanks

 

[Andu]

I'm not sure if anybody remembers but I was here before with trouble regarding my pppoe connection. A suggestion was to replace my old router that was serving as a stand in modem with a proper DSL modem. So I got the DSL-321B by D-Link which was suggested and unfortunately it didn't help. To be sure I tried to have the router establish the connection via PPPoE while the DSL modem was in bridge mode, as well as setting the modem to PPPoE mode and the N66U to Automatic IP as an alternative.
Yet still every day when the time for the forced reconnect comes the N66U has trouble handling the situation. What is concerning me is that this is even a problem when the modem does the work of reconnecting to the ISP.

The only improvement I can see over the situation before is that if I go to Network Management -> Connected/Disconnected and turn WAN off and on again it actually detects the change and the internet is accessible again.

I think there was a suggestion to add a cron job but I feel like that's just a workaround. Either my router is defective or there is an inherent problem in the firmware. And if my mind isn't failing me I think that in the earlier firmwares that I was using 108, 112 that there wasn't such an issue.

Just to be sure I will try one last alternative idea that I have. This is really unfortunate because otherwise I find the router rather brilliant but this really is a daily hassle which becomes even worse by the fact that there are people in my house that couldn't fix the problem by themselves if I'm not around and it's really annoying too.

 

[RMerlin]

What I've been working on lately...

 

[uncle.f]

What is concerning me is that this is even a problem when the modem does the work of reconnecting to the ISP.
...
Either my router is defective or there is an inherent problem in the firmware.

If you say that even your modem alone has a problem of re-connecting, how can it be a router defect? Sounds more like an ISP issue.

Anyway, if the time of re-connection is always the same, why not the cronjob workaround? As long as you are not running special apps directly on the router, should be no big deal to restart WAN connection or even reboot it routinely.

And if my mind isn't failing me I think that in the earlier firmwares that I was using 108, 112 that there wasn't such an issue.

Why not checking the old version then? If you can confirm that there is no issue with the earlier firmware, then somebody could potentially start helping. Although, I highly doubt that the behaviour will be different. Something tells me pppd daemon hasn't changed much between the releases.

 

[uncle.f]

Nfs

Question to RMerlin,

Currently /proc/filesystems shows support for squashfs, jffs2, ext2/3, vfat and ufsd.

Would it be possible to include support for NFS and maybe also CIFS into the next build's kernel? I would like to be able to access various file shares in my home network from the router.

Not sure if you recompiling kernel for your builds but if so, it should be trivial to enable some kernel options to support those filesystems.

 

[flar]

I did a search and didn't see anything on this, so I'll pass along an experience I had with this firmware in case it helps someone else.

I flashed the firmware from the official 178 and when I started back up I couldn't log in to the router to set it up - I kept getting the "must logout another client before logging in" error page. Even worse, I couldn't browse the web to find an answer because any web page I went to visit brought me the same error message.

This turned out to be what looks like an unintended interaction of 2 mechanisms here. The first mechanism redirects anyone who browses any web site after upgrading to the configuration wizard page. I suppose that makes it easy to get bootstrapped, but it can cause problems. The second issue is that once you land on that "click next to get started page" it considers you "logged in" and nobody else can do anything until you complete the initial setup.

The reason that causes problems is that networks are full of devices that love to "phone home" and may end up doing so immediately upon network service being restored (this is, admittedly, speculation on why my PS3 was the device listed as logged in whenever I tried to reboot the router). So they trigger the wizard page and they can do it before you can get logged in yourself on a computer. I finally tracked down that it was my PS3 beating me to the line and turned it off and I was able to get in and configure the wizard (after a power cycle on the router), but I think this feature could be tuned to avoid that Catch 22 - basically don't consider someone logged in until they actually start interacting with the wizard start web page. That way if I have a device somewhere on the network that managed to trip that page before I got a web browser pointed at it I can still complete the wizard steps even though I'm not the first on the page. I doubt my PS3 was smart enough to trigger the link attached to the "Next" button. It most likely sat and stared at the wizard start page.

On another note, I'd rather not have this feature, but as I said I can see how it helps a naive user get going. Perhaps it could be turned off by default on these 3rd party builds since we are all pretty savvy about router configuration if we're flashing a 3rd party firmware. The reason I am wary of it is that the devices which might "phone home" and get a wizard page instead may not recover gracefully and it may not be obvious that the device needs a reboot. It would be nice if the router simply came up in "conservative all-DHCP/NAT" mode and waited for the first true hit on .1.1 to trigger the wizard. Or, if it could somehow detect an actual web browser rather than a device doing a background network transaction that could help avoid sending a useless "start the wizard" page to an automated network device.

I'm not sure exactly what network transaction triggered it, but the device in question was my PS3 (as verified by MAC address) - I leave it running for Folding@Home and perhaps it pings Sony or the F@H servers when it sees the network restored?

 

[flar]

Sorry for that last really long post without saying a grand "THANK YOU" for putting out this firmware. I was getting really annoyed at a dozen or so little bugs in the official firmware that they manage to let slide release after release, so this was a breath of fresh air (and the first time in over a decade of using any of a dozen or so wireless routers that I finally decided to flash a 3rd party firmware ...

 

[flar]

And, now that I got the thank you out of the way, I have a nit with the stock firmware that also affects the Merlin builds that I'm hoping can be easily fixed.

I'm a fan of manual DHCP reservations (administered centrally through the router's reservation system rather than hard-coding the clients), so I've become intimately familiar with the N66U's address assignment page.

And, when I was configuring the Merlin firmware I made the same aggravating mistake I'd made 3 or 4 times since I got the router. I filled all the address assignments out for my network and the list was right there on the page all looking pretty, and then I went to another page because, after all, I had finished entering the assigments, right? Wrong - the page requires you to hit the Apply button when you are done, but it doesn't really have any visible warnings to this end, *AND* it lets you navigate away and lose all of your hard work without so much as a peep. GRRRR!? Would it be hard to add a navigation guard to stop and warn the user when they leave the page without applying changes?

On a related note - I absolutely love the reverse OUI lookups that you added to the main page in the "Client Status" list. Would it be hard to make them also available on the manual address assignment page? It would save having to have 2 windows open to try to identify the network clients...

 

[Andu]

If you say that even your modem alone has a problem of re-connecting, how can it be a router defect? Sounds more like an ISP issue.

That's not what I said. I said the router has difficulties detecting the change. The modem reconnects to the ISP just fine. I even confirmed this by looking into the modem's WebUI.

Anyway, if the time of re-connection is always the same, why not the cronjob workaround? As long as you are not running special apps directly on the router, should be no big deal to restart WAN connection or even reboot it routinely.

I guess if my last tries don't achieve the desired results and trying an old firmware doesn't help I'll have to set the cronjob. I'll just need some instructions where to place the code since I've never actually done this before.

 

[RMerlin]

Question to RMerlin,

Currently /proc/filesystems shows support for squashfs, jffs2, ext2/3, vfat and ufsd.

Would it be possible to include support for NFS and maybe also CIFS into the next build's kernel? I would like to be able to access various file shares in my home network from the router.

Not sure if you recompiling kernel for your builds but if so, it should be trivial to enable some kernel options to support those filesystems.

CIFS: Already there since 144.10.
NFS: Currently on the ToDO list.

 

[RMerlin]

@Flar:

Totally disabling EZ-QIS is something I'd rather not do, as it's one of my design goals to never take away any feature already there. I'd rather see Asus come up with a plan to fix this to avoid the scenario you mention. This could probably be done by allowing multiple users to hit the QIS wizard at the same time, and only consider a visitor "logged" once they hit the first "Next" button.

OUI lookups: I could add it to already existing DHCP entries, but then the Name field I added should normally make that lookup unnecessary. And doing lookups in the input field can't really be done in a clean way.

Preventing someone from leaving a page when they have unsaved changes can be done, but would require a LOT of changes to every single pages - that would make it a nightmare for me whenever re-basing on a new Asus release, so I'd rather avoid doing that.

 

[Andu]

So to end my story I finally managed to actually grab log info and Network Map info before and after power cycling. Maybe there is something useful in it or maybe not. I'll try an older firmware on the weekend before finding out and fiddling with cronjobs.

I'll just attach it as txt file so not everybody has to bother with it. I did however cover the IPs and Mac addresses.

 

[RMerlin]

So to end my story I finally managed to actually grab log info and Network Map info before and after power cycling. Maybe there is something useful in it or maybe not. I'll try an older firmware on the weekend before finding out and fiddling with cronjobs.

I'll just attach it as txt file so not everybody has to bother with it. I did however cover the IPs and Mac addresses.

These logs provide no usable information unfortunately.

I would suggest to ensure that you are running build 178 (either mine or Asus), and next time the connection stops working, check syslog for anything related to either pppd or rp-pppoe (see if there's an error or anything). Then, try turning the WAN off then back ON (through the main page), and see if the Internet works again. If it doesn't, then look again at the log to see what error message it gives you.

 

[thexile]

And, now that I got the thank you out of the way, I have a nit with the stock firmware that also affects the Merlin builds that I'm hoping can be easily fixed.

I'm a fan of manual DHCP reservations (administered centrally through the router's reservation system rather than hard-coding the clients), so I've become intimately familiar with the N66U's address assignment page.

And, when I was configuring the Merlin firmware I made the same aggravating mistake I'd made 3 or 4 times since I got the router. I filled all the address assignments out for my network and the list was right there on the page all looking pretty, and then I went to another page because, after all, I had finished entering the assigments, right? Wrong - the page requires you to hit the Apply button when you are done, but it doesn't really have any visible warnings to this end, *AND* it lets you navigate away and lose all of your hard work without so much as a peep. GRRRR!? Would it be hard to add a navigation guard to stop and warn the user when they leave the page without applying changes?

This is a clearly a user problem...

 

[RogerSC]

This is a clearly a user problem...

Yes, it is the user's problem when they lose a lot of time due to being distracted, perhaps make a typo, or are just not thinking at the wrong time. However, if there's a way to prevent that (like the "do you really want to delete that?" kind of thing), then given that software has lost a lot of the size and speed constraints that used to be the excuse for not helping the user be more productive, there's very little excuse for not doing it now.

On the other hand, I understand Merlin's argument that he's not in control of the software, he follows what Asus does. So this is really Asus's responsibility to take care of.

 

[RMerlin]

This thread has become far too long, making it quite unmanageable.

Tim has created a sub-forum dedicated to Asuswrt-Merlin, so we can have separate threads, making it far more easy for people to look for info or jump in conversations. Please start new individual threads, I will lock down this old thread in a few days.

 

[uncle.f]

CIFS: Already there since 144.10.
NFS: Currently on the ToDO list.

Thanks!

I actually found cifs.ko right after I wrote my previous post.
Unfortunately, I was unsuccessful in trying to map any CIFS share.

I have another Linux machine in my network and CIFS mount command that works on that box fails on the router.

The command that works on a Linux PC is

mount -t cifs -o username=guest //x.x.x.x/share /tmp/xxx

Same command fails on the router with the following message:

mount: mounting //x.x.x.x/share on /tmp/xxx failed: Invalid argument

(where x.x.x.x is an IP address, /tmp/xxx exists and the same anonymous share can be mapped from another Linux pc)

Maybe I should open new topic about it in the new sub-forum...

 

[RMerlin]

Thanks!

I actually found cifs.ko right after I wrote my previous post.
Unfortunately, I was unsuccessful in trying to map any CIFS share.

I have another Linux machine in my network and CIFS mount command that works on that box fails on the router.

The command that works on a Linux PC is



Same command fails on the router with the following message:



(where x.x.x.x is an IP address, /tmp/xxx exists and the same anonymous share can be mapped from another Linux pc)

Maybe I should open new topic about it in the new sub-forum...

Taken from the firmware documentation:

* Mounting remote CIFS shares on the router *
You can mount remote SMB shares on your router. The syntax will
be something like this:

mount \\\\192.168.1.100\\ShareName /cifs1 -t cifs -o "username=User,password=Pass"

(backslashes must be doubled.)

See if you get better luck with this.