Custom firmware build for R7800 v. 1.0.2.31SF

[mr.fabe]

Strange. Maybe there were some temporary problems on Netgear cloud server.

Would be good if mr.fabe check it now.

Voxel.

I am currently using a modified version of .28 that fixed my Readycloud registration issue. If it were a Readycloud system issue, I don't believe the Netgear folks would have given me a patched rom but I've been wrong before

The main reason I use Readycloud is to provide limited expiring access to files for other client users. I could just allow FTP access to my servers, however, the Readycloud share features are pretty convenient to me.

I'll give the Voxel .31 another run...

 

[Voxel]

Info for Entware-3x users:

New version of Entware-3x for Cortex-A15 architecture is released (170602). More than 90 packages are upgraded.

Good news for those who are using R7800 as a webserver. Zyxmon made a test version for MariaDB:

Mysql 5.1 replaced by Mariadb 5.5.56


The links are:

http://ownweb.eu/Downloads/Entware/
(thanks to vladlenas for his help and hosting). Automatic installation tuned to R7800 with my build.

or

http://cortex-a15.zyxmon.org/binaries/cortex-a15-3x/
(thanks to zyxmon for hosting). Classic Entware-3x installation known for Entware users.

It is enough to do "opkg update" and "opkg upgrade" if you did not use Mysql. Otherwise, remove Mysql packages first.

Voxel

 

[ulaganath]

It is troublesome to maintain and to release two versions of firmware, when second is created especially for you. I do not think that rest users of my custom firmware would be happy if I’ll replace QoS DB by older version, because practically there are no feedbacks from others that they have problems with latest QoS DB.

Let’s do the following: I put previous QoS DB packages to public downloads:

http://ownweb.eu/Downloads/R7800-Voxel-firmware/QoS DB/

You can extract *.ipk files from ZIP:

macoui-db_20160607112939-cloud-gb8be08c-generic-1_all.ipk
p0f-db_20160607112934-cloud-gbaabd85-generic-1_all.ipk
policy-redis_20160607112945-cloud-g3bc8341-generic-1_all.ipk
wopr-db_20160607112951-cloud-g0e9656a-generic-1_all.ipk

and install them manually from telnet/ssh to your router using “/bin/opkg” command. Use full path to /bin/opkg command, to avoid call of Entware’s opkg (/opt/bin/opkg).

This program “/bin/opkg” has an option “--force-downgrade” allowing to downgrade installed package. Local files could be installed too. Something like:

/bin/opkg --force-downgrade install macoui-db_20160607112939-cloud-gb8be08c-generic-1_all.ipk

NOTE: I did not test this with my router, because I do not want to downgrade QoS DB. But it should work.

Voxel.

Hi Voxel,

Wanted to update on your solution. it worked like a charm.

I was in 30F with v2. No problem at all. The QOS never crashed and worked with day in out of disconnected clients and it was running cool. So it was time for me to test your solution.

I just flashed without reseting the settings and QOS was iniitally working next day morning resuming the PC from hibernate the browser as usual prompted for the credential. Booom the QOS broken.

So i went to downgrade the files you shared and though of restarting the streamboost can load the new files. But it seems the same error. Instead of stopping the services it start and then stop.

Hence i had rebooted the router to see if the new DB worked.

It worked after reboot as usual for even v4.0. Then i went out the PC in hibernate mode for 2 hours came back and booted the pc and Same process. Bingo the QOS worked like charm.

The reason i curious on this is that. It does well only when the QOS is alive. As the priority i set is saved and served well. When it crashed even 2-3 devices active casuing delay.

Now it seems the v4.0 DB needs a rework again. What went wrong is upto Netgear and Chip manufacture.

 

[Killhippie]

Hi Voxel,


Now it seems the v4.0 DB needs a rework again. What went wrong is up to Netgear and Chip manufacture.

I have no issue at all with the new database and have seen no complaints on here apart from yours with respect. I think the issue is more to do with your network and router setup possibly. If there were problems the Netgear forums would be very noisy with people complaining, and they are not, the same with this forum also. I'm glad you got things to work out but the latest database really should work work fine. Also how do you mean "boom the OoS was broken" when you opened a browser? QoS operates in the background and it does not bring the router down so I really don't see what your issue is, or what you think QoS is actually doing that's 'broken'

 

[ulaganath]

As i keep explaining. You didnot see that because you didnt broke it yet. Here is how i did it most of the time.

I keep the browser open in router gui page on QOS attached devices. For active traffic the page give + symbol to see the applicationwise traffic. say youtube, netflix, dropbox etc.

It works well during the initial boot. .

Once i put my pc to sleep hibernate. * with the browser open at the same page. As for the security reason 10-15 mintues of inactivity the browser session will be closed by router which needs re-authentication. Let say after 1hr i resume the pc and the browser is active which tries to pull the fresh data from the browser.

Router needs authentication. i save the credentials through lastpass. So i give it instantly instead of typing.

Boom the + Symbols are gone and not working anymore The saved priority over time will be reset to low or anything.

I even tried to reboot the streamboost. No luck it stops instantly

Only way to make it work is reboot and it wont even last for 2-3 hours.

I just said one way it is crashing. But there are many ways for me it crashed. But with old DB even with 30F i had no crash and now with 31F with old DB its working as i want it be .

 

[ulaganath]

with 100% ISP speed of game download from Origin client the cpu is spiking to 70% with +/- 15% flucutation

Is this normal.

As what if i upgrade fo 200Mb or 1gb . Can this router handle it. It is fastest of quadband series except x10 which is way beyond.

I monitored through netdata app through entware.

it is not peak or fluctuted max. It even went 75

I am glad i am getting full throughput. Only clients like IDM, Steam and orgin can pull such a constant speed. More over my ISP also never throtles gives rated speed most of the time.

 

[Killhippie]

As i keep explaining. You didnot see that because you didnt broke it yet. Here is how i did it most of the time.

I keep the browser open in router gui page on QOS attached devices. For active traffic the page give + symbol to see the applicationwise traffic. say youtube, netflix, dropbox etc.

It works well during the initial boot. .

Once i put my pc to sleep hibernate. * with the browser open at the same page. As for the security reason 10-15 mintues of inactivity the browser session will be closed by router which needs re-authentication. Let say after 1hr i resume the pc and the browser is active which tries to pull the fresh data from the browser.

Router needs authentication. i save the credentials through lastpass. So i give it instantly instead of typing.

Boom the + Symbols are gone and not working anymore The saved priority over time will be reset to low or anything.

I even tried to reboot the streamboost. No luck it stops instantly

Only way to make it work is reboot and it wont even last for 2-3 hours.

I just said one way it is crashing. But there are many ways for me it crashed. But with old DB even with 30F i had no crash and now with 31F with old DB its working as i want it be .

Ok for s start you are relying on the GUI far to much, it does not need to be open, also I don't suffer any of your issues, the + symbol is always there, my priorities are always as I leave them. Mostly I let the QoS decide on priority because it gets it right, it knows my PS4 Pro needs a higher priority than my iMac and my printer needs less than my wireless speakers and that if I'm watching Netflix that it will make sure that stream runs well while say my PS4 or iPad/iPhone etc are maybe updating in the background.

You should let the QoS do its thing and not muck about with it to much, its designed to be fairly autonomous, maybe you need to change a few things for personal preference hence the GUI to access that. My router runs day and night and the QoS works fine all the time and the QoS never 'crashes' you are talking about the GUI being buggy not the actual QoS itself really. It does feels like you are putting to much emphasis on the GUI maybe? The only time the router may reset the QoS from the changes I have made is it its turned off with no power. Your pc hibernating should have no effect on the routers QoS as the priorities are set whether the pc is asleep or not in the router. My iMac sleeps, wakes and all is fine and the QoS if I log in is doing what it is meant to do.

 

[ulaganath]

thats true for when it is working fine. I dont rely on GUI. but i do know when it broke crash . I mean it. It doesnt work the way when it is crashed. Only a reboot make it to work.

For now i am fine with old DB as there is not much with new DB or any specific related to change notes from either party.

Was this cpu consumption normal for even 100Mbps continous transmission.

i cant think of 200 or more speed will do this router. I know it can handle . But wondering too much of cpu consumption lead to heating which is why i am concerned.

 

[Killhippie]

thats true for when it is working fine. I dont rely on GUI. but i do know when it broke crash . I mean it. It doesnt work the way when it is crashed. Only a reboot make it to work.

For now i am fine with old DB as there is not much with new DB or any specific related to change notes from either party.

Was this cpu consumption normal for even 100Mbps continous transmission.

i cant think of 200 or more speed will do this router. I know it can handle . But wondering too much of cpu consumption lead to heating which is why i am concerned.

Once again I can only speak from my perspective, I have never had to reboot the router due to QoS crashing simply because it hasn't crashed. Its possible rebooting may alter the QoS I guess. I think chopping and changing firmware so much is probably not the best thing to do otherwise you could I suppose end up with corrupted nvram. As to overheating I would not worry, Qualcomm chipsets run cooler than Broadcom ones from what I have read and experienced. Mine with 4K HDR streams 11 wifi connections and Ethernet ones etc never really gets warm enough to be of any concern. I think you should relax and not worry to much about that

 

[Charles]

Voxel,

I want to update the pppd version from the ancient 2.4.3 to at least 2.4.6, in order to have support for RFC 4638 and ppp-max-payload to allow a full 1500 MTU to be used for PPPoE connections. I already have this on a secondary R7800 running LEDE.

In your firmware, would it simply be a matter of replacing the ../git_home/ppp.git directory with the latest pppd sources, and then recompiling?

Thanks

 

[Voxel]

Voxel,

I want to update the pppd version from the ancient 2.4.3 to at least 2.4.6, in order to have support for RFC 4638 and ppp-max-payload to allow a full 1500 MTU to be used for PPPoE connections. I already have this on a secondary R7800 running LEDE.

In your firmware, would it simply be a matter of replacing the ../git_home/ppp.git directory with the latest pppd sources, and then recompiling?

Thanks

Charles,

Not so easy. NETGEAR developers keep in /git_home not original sources of packages, but modified by them. These modification could be used for internal need in firmware, something like use of files in /tmp with specific info needed for other modified packages, scripts or prebuilt binaries. So you should do comparison of this ppp.git with original ppp 2.4.3, find changes and integrate with other 2.4.6. For example proftpd.git version 1.3.3 in GPL of stock firmware contains some specific changes, and when I upgraded proftpd to 1.3.5b I had to find these changes (original 1.3.3 vs 1.3.3 from git_home) and to use them in 1.3.5b. Otherwise you may damage other functionality...

Voxel.

 

[Voxel]

Charles,

Not so easy. NETGEAR developers keep in /git_home not original sources of packages, but modified by them. These modification could be used for internal need in firmware, something like use of files in /tmp with specific info needed for other modified packages, scripts or prebuilt binaries. So you should do comparison of this ppp.git with original ppp 2.4.3, find changes and integrate with other 2.4.6. For example proftpd.git version 1.3.3 in GPL of stock firmware contains some specific changes, and when I upgraded proftpd to 1.3.5b I had to find these changes (original 1.3.3 vs 1.3.3 from git_home) and to use them in 1.3.5b. Otherwise you may damage other functionality...

Voxel.

Forgot to say: this process requires knowledge of C or C++ and basic understanding logic of program. And maybe guessing what for these changes (what other packages need from these changes), some kind of debugging...

Voxel.

 

[Voxel]

For now i am fine with old DB as there is not much with new DB or any specific related to change notes from either party.

I did not quite understand: i.e. you flashed 1.0.2.31SF and downgraded QoS? And now it is OK for you. Right?

Voxel.

 

[ulaganath]

exactly. I flashed from 30F v2.0 db to 31f v4.0 i believe. It was working fine as usual. After 8 hours of resuming from hibernated the re-authencation crashed the QOS streamboost.

Manually restart of streamboost didnt work. it was as usual weired starting the process instead of stopping and then starting.

I did downgrade the qos files you shared and i still couldnot restart he streamboost. But i did a reboot whola. it works

So far System Up Time 01:05:34:31

Did multiple re-authentication after hibernate resume. It was working as usual.

Just a 80kg bunch of files decides the qos process stability. So much .

Hoping the next version wont crash . It might take ages i know.


Coming to the CPU usage. Is that normal for 100mbps to hit 70%. average was 61%

Also usb host read and write of 2.0 port is hitting close to 80% . I know it takes cpu hit a lot. But i didnt affect my download speed. But at time the webpage request is slow.

 

[ulaganath]

Hi Voxel,

I faced a weired issue. The vpn server setup allow me to connect to the router as client with different subnet at the 3rd level.

but when i hit the router gui it says

404
this server doesnot support the operation requested by the client

But i can connect to the client which is another router in bridge wired mode via lan port . I can hit the gui which is LUCI LEDE firmware on WNDR4720.

So what am i missing here.

Also the UDP alone works for the mobile TCP fails. May be for TCP i need to refresh the certiface.

As for UDP it works for both normal and 2x client series.

Let me know why it is not visible for the r7800 but working for bridged router

 

[mr.fabe]

I am currently using a modified version of .28 that fixed my Readycloud registration issue. If it were a Readycloud system issue, I don't believe the Netgear folks would have given me a patched rom but I've been wrong before

The main reason I use Readycloud is to provide limited expiring access to files for other client users. I could just allow FTP access to my servers, however, the Readycloud share features are pretty convenient to me.

I'll give the Voxel .31 another run...

Well, I've been running the Voxel .31 and I am happy to report that I was wrong . All the features of Readycloud work as intended in this version. It could be due to me not clearing the NVRAM, but I don't really care since Readycloud has become a non-issue for me. In addition, I have had no problems with QoS and VPN... both are working very well. Thank you Voxel... love your hard work on the the firmware.

 

[Voxel]

Well, I've been running the Voxel .31 and I am happy to report that I was wrong . All the features of Readycloud work as intended in this version. It could be due to me not clearing the NVRAM, but I don't really care since Readycloud has become a non-issue for me. In addition, I have had no problems with QoS and VPN... both are working very well. Thank you Voxel... love your hard work on the the firmware.

Just probably there were temporary problems with NG's ReadeSHARE site.

Voxel.

 

[Voxel]

I faced a weired issue. The vpn server setup allow me to connect to the router as client with different subnet at the 3rd level.

but when i hit the router gui it says

404
this server doesnot support the operation requested by the client

But i can connect to the client which is another router in bridge wired mode via lan port . I can hit the gui which is LUCI LEDE firmware on WNDR4720.

So what am i missing here.

Also the UDP alone works for the mobile TCP fails. May be for TCP i need to refresh the certiface.

As for UDP it works for both normal and 2x client series.

Let me know why it is not visible for the r7800 but working for bridged router

There was similar report:

https://www.snbforums.com/threads/c...-2-25sf-v-1-0-2-29sf.38032/page-2#post-315135

Please check my answer too.

Voxel.

 

[ulaganath]

Okay i got it. Works only on one mode.

How do i select that mode. It just give a config file and i import to smartphone. I dont know if it gives option to connect either mode.

 

[Voxel]

Okay i got it. Works only on one mode.

How do i select that mode. It just give a config file and i import to smartphone. I dont know if it gives option to connect either mode.

Smartphone config downloaded from router's GUI is using TUN. So no way. Or seek for TAP OpenVPN client and modify your client config for TAP. I know one guy is using TAP for his Android with my build.

Voxel.