[378.56.2] Merlin keeps asking for password, can't change any settings

[RMerlin]

Your suggestion about a long or invalid hostname is very likely. When I click the network map / view list sometimes it briefly shows the list (less than a second) before crashing and there is one name that is very long (that seems to have been automatically added) but there is never enough time to edit it. Is there a file I can edit via SSH to correct the long/invalid hostname?

Easiest way to troubleshoot this would be to start with a minimal number of clients connected. Then, add more clients until you can track down which one is causing it. It will be easier then to start figuring out what debugging could be done next. Without a starting point, a debug build cannot just be compiled. Unlike with a full-fledged Linux environment, the router does not generate core dumps when an application crashes.

 

[effadj]

Check the System Log. If you see a "restart_httpd" from the "atchdog" process, then it means something caused the web server to crash and to be restarted.

Yup....

Apr 19 20:13:04 watchdog: restart httpd
Apr 19 20:13:04 rc_service: watchdog 250:notify_rc start_httpd
Apr 19 20:13:05 httpd: start httpd​

 

[effadj]

Easiest way to troubleshoot this would be to start with a minimal number of clients connected. Then, add more clients until you can track down which one is causing it. It will be easier then to start figuring out what debugging could be done next. Without a starting point, a debug build cannot just be compiled. Unlike with a full-fledged Linux environment, the router does not generate core dumps when an application crashes.

Can you say what the length limit and valid characters of a host name can be?

 

[RMerlin]

Can you say what the length limit and valid characters of a host name can be?

16 characters, only alphanumeric and dashes.

Longer names are valid, however in older firmware releases they could causes crashed in networkmap. It's still possible that there might be some areas of the firmware that still fail to properly validate the length when manipulating them.

 

[effadj]

16 characters, only alphanumeric and dashes....

I used the router's NETSTAT tool to look at the TCP sockets. I see this within:

tcp 194 0 router.asus.com:46072 PlayFi-Family_Room00A5C8:49152 CLOSE_WAIT​

I have a name (PlayFi-Family_Room00A5C8) from wireless speakers hub that uses underscores in them (_). I can not control that name. When viewed under the client list within the Network Map tab, that hostname is displayed as "Play-Fi_Device". Could this be the problem?

 

[RMerlin]

I used the router's NETSTAT tool to look at the TCP sockets. I see this within:

tcp 194 0 router.asus.com:46072 PlayFi-Family_Room00A5C8:49152 CLOSE_WAIT​

I have a name (PlayFi-Family_Room00A5C8) from wireless speakers hub that uses underscores in them (_). I can not control that name. When viewed under the client list within the Network Map tab, that hostname is displayed as "Play-Fi_Device". Could this be the problem?

Underscores are frowned upon (because they are reserved characters when a hostname is part of a valid DNS zone), but they should not create any problem within a home LAN.

 

[JamesG]

I have some positive results with only a half implemented plan, though I don't understand the reason for it yet.

I had already identified a hostname that looked like it might be causing issues based on previous suggestions: Wistron InfoComm(Zhongshan) Corpor (truncated because it was so darn long it didn't fit the screen.) This is actually an RD340 Lenovo Server management interface... go figure.

I enabled JFFS and was going to try one of the suggestions earlier in this thread to add this to /jffs/configs/dnsmasq.conf.add, but just the act of enabling JFFS seems to have stopped the httpd crash when viewing the hosts from the Network Map / View List page. I was also successfully updated the hostname to change Wistron... to something more descriptive and compliant.

There were no other changes in the network since I can't turn off that machine running some VMs (assuming it was the culprit.) While this worked, it is not clear how enabling JFFS would fix this problem. For the moment, no more crashes of httpd. To add to this, one of the other RT-N66Us that I have that didn't exhibit this crash also had JFFS enabled. Again, this may just be a coincidence so YMMV.

 

[RMerlin]

I enabled JFFS

How was it disabled? It should never be disabled anymore, the option to do so was removed many months ago. Asuswrt stores a lot of things in JFFS now, including the networkmap database and the OpenVPN key/certs.

If it was disabled back when the option was still available, then it means you probably skipped the mandatory factory default reset at that time. That would certainly explain your random httpd crashes, as it needed to be able to access the JFFS partition.

 

[JamesG]

Thank you Merlin, that explains it quite clearly, and not a coincidence. I had upgraded three routers at the same time and intended to factory reset them all. Either the interface crashed when I was attempting to reset or I just plain missed resetting it. I just needed to figure out what was different between them

The input is much appreciated. Hopefully this saves someone else some trouble.

 

[effadj]

I enabled JFFS and was going to try one of the suggestions earlier in this thread to add this to /jffs/configs/dnsmasq.conf.add, but just the act of enabling JFFS seems to have stopped the httpd crash when viewing the hosts from the Network Map / View List page.

Where is that setting? I don't see it. The only settings under "Administration" are:

Format JFFS partition at next boot
Enable JFFS custom scripts and configs​

Thanks!

 

[JamesG]

You have found the exact settings used to enable JFFS, but if you are running a recent firmware it should already be enabled after doing the mandatory factory reset that Merlin indicated. If it is not, a factory reset should enable it.

Since the factory reset didn't happen on mine (didn't work or missed doing it) I originally enabled it by setting "Enable JFFS Custom scripts..." set to Yes and the "Format JFFS partition at next boot" to Yes since the partition didn't exist and wasn't formatted before.

If you have already enabled JFFS, you probably don't want to do this since it will wipe out any configs stored there. You can check by SSHing in and seeing what is mounted, or just change to the directory and see if there are some subdirectories for configs, scripts, openvpn and usericon. If you don't have JFFS, probably doing a factory reset is not a bad idea, but someone more knowledgeable can pipe in here.