[Fork] Asuswrt-Merlin 374.43 LTS releases (Archive)

[john9527]

Excellent job, John. Version 7 rock solid so far. Not sure if it is just fork update placebo effect but I notice it seems to reconnect faster when I wake my computers.

Placebo effect.....I like that

Nothing knowingly done that would have made a change on reconnects....

 

[john9527]

Not sure what changed in your last update, but my webpages now load much faster

Link speed is the same as before but pages definitely load faster

Wish I had a explanation for you, but I'm at a loss. Nothing specific that I did.

But, when I had my compile difficulties, I made an offering to the router gods....maybe they smiled upon us

EDIT: I did forget to ask....are you running an OpenVPN client on the router? If so, you probably got at 30-50% throughtput improvement just by loading the code.

 

[john9527]

There was no need to add the fix in the final build as all this was already done hence final release.

Not exactly....the DSCP fix still needs to be enabled. I just added it into the gui as an option in the Firewall page. I'm guessing that there may be some cases where it wouldn't be a good thing to have enabled permanently, maybe if you were using VOIP or IPTV.

 

[Deleted member 22229]

Not exactly....the DSCP fix still needs to be enabled. I just added it into the gui as an option in the Firewall page. I'm guessing that there may be some cases where it wouldn't be a good thing to have enabled permanently, maybe if you were using VOIP or IPTV.

Ok i stand corrected i was not aware of that. Thanks for making that clear.

 

[johtaja]

If you did that, then your modem would also have to act as a DHCP server to give a separate address to the Obi and the router. (I think most modems can do this or be set to do this). Then your other other clients connected to the router and the router would be in the Double-NAT connection. I don't have any experience with double-NAT setups, but I think it would end up complicating your life more than you want For example, I think the OpenVPN clients have trouble dealing with double-NAT, so you may just end up changing one problem for another.

If anyone else has more experience here....please chime in....

Here in finland, at least my ISP (TeliaSonera), gives max 5 public IPs per customer, so if that's the case with Lotta Cox's ISP, it should be possible to connect as Lotta Cox had in mind, and keep modem bridged.

 

[soundping]

Minor issue.

Wireless - Visible Networks (Site Survey)

All Radio strength icons are displayed at full.


Correction: They seemed to fixed themselves.


Nope Now they are 100% again and I'm missing about seven people on the list.

 

[john9527]

Minor issue.

Wireless - Visible Networks (Site Survey)

All Radio strength icons are displayed at full.


Correction: They seemed to fixed themselves.

I was just going to ask if you were in 'hot zone' since they were working fine for me.

 

[soundping]

Did a factory reset and now they look normal.

 

[Jaded]

...noticed a bug... Could explain kindle issues if other user...

Was having issues with my rt-68u losing connection to my bridge every 10-20mins ... Wouldn't re-authenticate... Would drop until I would manually try to connect to bridge....

Went into wireless settings and noticed the beamforming on my 5ghz (both beaforming options) were turned back to on after upgrading to build 7... Turned off and haven't had a disconnection in 30 mins.... Looks promising...

Check your beamforming settings if you are having WLAN issues...

...update....
Been an hour and authenticating normally now with beamforming disabled again.

 

[RMerlin]

Wish I had a explanation for you, but I'm at a loss. Nothing specific that I did.

My guess is, it's the reboot involved with the firmware upgrade that did it. I've seen plenty of these placebo posts myself over the years...

 

[john9527]

Was having issues with my rt-68u losing connection to my bridge every 10-20mins ... Wouldn't re-authenticate... Would drop until I would manually try to connect to bridge....

Went into wireless settings and noticed the beamforming on my 5ghz (both beaforming options) were turned back to on after upgrading to build 7... Turned off and haven't had a disconnection in 30 mins.... Looks promising...

Check your beamforming settings if you are having WLAN issues...

...update....
Been an hour and authenticating normally now with beamforming disabled again.

Well, I'll be damned....good catch. Turns out when you do a code load, the router does sort of a mini reset. It resets the memory size, wipes out the info on usb connections so everything get re-detected/initialized on the new code, things like that. Buried in that sequence, for ARM routers, they turned all the beamforming options on if the 'extended' code id (anything after the '_') had changed. So, going from 374.42 to 374.43 wouldn't turn beamforming on, but 374.43_1 to 374.43_2 would?????? And since my code id is in the extended area beamforming will get turned back on. Been there since Merlin's original code.

Congratulations on finding the first commit for the next release

In the meantime, those with ARM routers (AC56, AC68) will need to manually reset the Beamforming options if they changed them from the defaults (all enabled). I've also added a 'KnownBugs' file to the download area.

Note to Merlin....this has already been removed from your master at some point.

 

[scaramonga]

So going by that, its recommended to have Beamforming off then?

 

[john9527]

So going by that, its recommended to have Beamforming off then?

No hard and fast rule for the Beamforming options....from what I've read they seem to help some and hurt others. It's very client and environment dependent.

 

[Jaded]

..another small bug

In the wireless clients tab under system log DHCP clients lose their host names after the dhcp refreshes after the 24 hour lease period expires


...you have to disconnect and reconnect clients to have host name reappear if you hit this and it bugs you.. Increase the dhcp lease is what I am using as a workaround... This will also repopulate your wireless tab with host names if you do this prior to reconnecting your client.... Max lease time is 604800 seconds..

 

[john9527]

In the wireless clients tab under system log DHCP clients lose their host names after the dhcp refreshes after the 24 hour lease period expires


...you have to disconnect and reconnect clients to have host name reappear if you hit this and it bugs you.. Increase the dhcp lease is what I am using as a workaround... This will also repopulate your wireless tab with host names if you do this prior to reconnecting your client.... Max lease time is 604800 seconds..

If I understand where you are looking correctly, under System Log/Wireless Log

I'm sorry, but I can't recreate any lost name problem. I set my DHCP lease time to 600 sec and went through 3 renew cycles without any problem on both 2.4 and 5GHz clients.

The only thing I can think of is if you have clients that are both wired and wireless connection. If both are active, it will wipe out the wireless name since you can't have two clients with the same name on the network at the same time.

EDIT: I think I may have figured it out. After I posted, one of the clients went to 'sleep'. The name blanked, the connected time went to all zeros, and the Flags changed from STAU to just ST. As soon as I woke up the client (Windows 7) everything came back.

 

[Jaded]

Yes, that is where I am seeing it.. However....No duplicate names.

...The clients dropped out at around the 24 hour mark. Noticed few hours after so and assumed it occurred at lease renew since all devices had dhcp req/offer/renews in the systems log with nothing else indicated.

Other observations...
..Their connection status indicated there was not a disconnection from the router
..devices were connected to both radios. (4 devices to 2.4 and 3 to 5ghz)
..making sure the device was awake and able to connect to Internet did not repopulate host names
..host names were repopulated once I disconnected and reconnected from router
..no other messages were in system log
..I keep an iPad connected to the managment page over the 2.4ghz radio and it was the only device which kept its host name (always do when I load new code)
..I use static dhcp mapping for all clients with host names configured
.. All devices were flagged as P A U. I confirmed not a sleep mode issue by accessing Internet. Names did not repopulate.
..I have multiple static ip devices that never show host name connected to both wireless radios


Saw your post .... Update...
I do notice when a device does sleep or I connect to a different wireless network that host name does go blank too. Thought that was expected behavior since device isn't technically connected..

Will post if it happens again.

 

[john9527]

@Jaded....check my post edit....we may have passed each other with posts.

 

[Jaded]

Can you test with a static ip device connected to each radio and see if occurs?

I wonder if that's the issue...


Also..it might matter...
My 5 ghz connected devices are actually connected over a. Wireless bridge... So they can never "go to sleep". Most of these devices are static ips with only 3 dhcp devices... No P flag on their connections even... STAU flags for all 5ghz devices... All of these dhcp devices lost names on top of the 2.4 ghz clients which do carry P flags

 

[TheImmortal]

Sorry if it's been discussed, but I've found no answers to this 'till now...

I've been constantly upgrading John's fork updates since the first release but every month or so my router starts failing and logs these:

"Jan 26 11:27:10 openvpn[30830]: Diffie-Hellman initialized with 512 bit key
Jan 26 11:27:10 openvpn[30830]: Cannot load private key file server.key: error:0B080074:x509 certificate routines:X509_check_private_key:key values mismatch
Jan 26 11:27:10 openvpn[30830]: Error: private key password verification failed
Jan 26 11:27:10 openvpn[30830]: Exiting due to fatal error"

I'm getting connection issues on all devices once that happens and the only way to get past this is to actually reboot the router. Anyone happens to have a workaround? I'm not really sure what SSL cert the error refers too, nor why would the mismatch occur.


Thank you.

 

[ika]

I was able to exactly recreate your scenario (that had been bugging me as well and I just hadn't taken the time to look at it).

Taking a quick look at the code, it looks like it's trying to pre-allocate the db directory prior to starting minidlna and doesn't properly take into account someone moving it. Actually, it looks like it tries but doesn't quite get there. So the answer right now is just to delete the directory.

I did a workaround with a minidlna.postconf script instead of replacing the conf file. I moved it to '.minidlna' so it would be hidden and not show up if you are also using the drive as an SMB share and then just followed up with a delete to the empty directory.

Hi,

Sorry I did not reply earlier (family matters:/), I read and thanked your post, but did not have the time to actually reply.

The hidden folder is a great idea, dunno how did I miss that, thank you very much. And deleting with a script is fine for me too ofc, don't worry about it, I was just curious if I made an error somewhere, so I asked for an advice as a second opinion.

Thanks again for your time and for your great work.
Ika