Voxel Custom firmware build for R7800 v. 1.0.2.62SF

[Voxel]

Continuation of
. . .
https://www.snbforums.com/threads/custom-firmware-build-for-r7800-v-1-0-2-61sf.49358/

New version of my custom firmware build: 1.0.2.62SF.

Changes (vs 1.0.2.61SF):

1. OpenSSL package is upgraded 1.0.2p->1.0.2q.
2. DNSCrypt Proxy v.2 (2.0.19) is included into firmware:

to enable DNSCrypt Proxy v.2 run the commands from telnet/ssh console:

nvram set dnscrypt2=1
nvram commit​

and reboot your router;
to disable DNSCrypt Proxy v.2 run the commands from telnet/ssh console:

nvram set dnscrypt2=0
nvram commit​

and reboot your router.​

3. ipset package and its dependences are added into firmware.
4. unbound package (used in stubby) is upgraded 1.8.1->1.8.3.
5. wget package is upgraded 1.19.5->1.20.
6. util-linux package is upgraded 2.32.1->2.33.
7. haveged package is upgraded 1.9.2->1.9.4.
8. ethtool package is upgaded 4.18->4.19.
9. libjpeg package is upgraded 9a->9c.
10. curl package is upgraded 7.61.1->7.62.0.
11. libgcrypt package is upgraded 1.8.3->1.8.4.
12. libubox package is upgraded 2018-07-25->2018-11-16.
13. sqlite package is upgraded 3240000->3250300.
14. jq package is upgraded 1.5->1.6.

The link is:

https://www.voxel-firmware.com (thanks to vladlenas for his help with hosting).

NOTES:

1. stubby (DNS-Over-TLS) is already included into firmware. To enable stubby run the commands from telnet/ssh console:

nvram set stubby=1
nvram commit​

and reboot your router; to disable stubby run the commands from telnet/ssh console:

nvram set stubby=0
nvram commit​

and reboot your router. Currently it is configured to use Cloudflare servers (DoT). You may customize your config file:

/etc/stubby/stubby.yml

2. DNSCrypt Proxy v.2 is already included into firmware. To enable DNSCrypt Proxy v.2 run the commands from telnet/ssh console:

nvram set dnscrypt2=1
nvram commit​

and reboot your router;
to disable DNSCrypt Proxy v.2 run the commands from telnet/ssh console:

nvram set dnscrypt2=0
nvram commit​

and reboot your router. Currently is is configured to use Cloudflare, ipredator, scaleway-fr, dnscrypt.eu-nl, dnscrypt.eu-dk. You may customize your config file /etc/dnscrypt-proxy-2.toml to add/remove your selected servers.

3. If both (stubby and DNSCrypt Proxy v.2) are enabled, stubby has highest priority (it will shutdown all dnscrypt (v1 and v2)). DNSCrypt Proxy v1 if still used has lowest priority.

Voxel.

 

[Gallo]

Props Voxel excellent work!!!
Was this issue addressed or is it something that will have to be manually done after every install.
https://www.snbforums.com/threads/c...r-r7800-v-1-0-2-60sf.48805/page-2#post-430750

 

[Voxel]

Was this issue addressed or is it something that will have to be manually done after every install.

It is better to have your own backup keeping for example /root/.ssh /root/.profile with PATH settings etc.

https://www.snbforums.com/threads/e...xel-routers-for-sky-uk-etc.41720/#post-353344

Could be restored from telnet after new flashing.

Voxel.

 

[EmeraldDeer]

In case anyone out there is:
1) Running an R7800 as an access point
2) About to run an upgrade to Voxel 62
3) Comfortable with telnetting in and running commands

Could you please check the dnsmasq command in top on the AP before and after the upgrade. After the 62 upgrade, my R7800 AP had dnsmasq pinning one of the CPU cores. I shut it down with /etc/init.d/dnsmasq stop. I do not know whether this had been happening on the AP beforehand with Voxel 61.

Dnsmasq on my R7800 router is OK as far as CPU usage. However, the default enabling of the dnsmasq option no-negcache is not appropriate in my situation. There are high numbers of negative reverse lookups from my LAN which flood my DNS provider unless I remove the no-negcache option from /etc/dnsmasq-resolv.conf and reboot.

 

[eliz82]

@Voxel I had Entware with a webserver (nginx+php+mysql) running for months on my ssd connected on esata port on my R7800. Yesterday without making any changes to my router, I suddenly observed webserver don't work properly, sda1 has become read-only. If i try to delete or create a file in "/opt" i get error.

rm /opt/www/default/index.html
rm: remove '/opt/www/default/index.html'? y
rm: cannot remove '/opt/www/default/index.html': Read-only file system
------------------
mount

[...]
/dev/sda1 on /tmp/mnt/optware type ext2 (ro,nodev,noatime)
------------------
dmesg | grep "sda"

Adding 2048572k swap on /mnt/sda1/swap.  Priority:-1 extents:518 across:2083840k SS
EXT2-fs (sda1): error: ext2_lookup: deleted inode referenced: 869517
EXT2-fs (sda1): error: remounting filesystem read-only
EXT2-fs (sda1): error: ext2_lookup: deleted inode referenced: 869518
EXT2-fs (sda1): error: remounting filesystem read-only
EXT2-fs (sda1): error: ext2_lookup: deleted inode referenced: 869519
EXT2-fs (sda1): error: remounting filesystem read-only
EXT2-fs (sda1): error: ext2_lookup: deleted inode referenced: 869520
EXT2-fs (sda1): error: remounting filesystem read-only

as you can see it clearly show my sda1 is mount read-only. and it seems it because of some errors. any idea what can it be ... ssd failure error, cable error, software error?

if i try an disk error check:

e2fsck -n /dev/sda1

e2fsck 1.43.4 (31-Jan-2017)
Warning!  /dev/sda1 is mounted.
optware contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
The bad block inode looks invalid.  Clear? no
Deleted inode 884806 has zero dtime.  Fix? no
Deleted inode 983421 has zero dtime.  Fix? no
Deleted inode 983635 has zero dtime.  Fix? no
Deleted inode 984792 has zero dtime.  Fix? no
Deleted inode 984794 has zero dtime.  Fix? no
Deleted inode 984927 has zero dtime.  Fix? no
Deleted inode 985956 has zero dtime.  Fix? no
Deleted inode 985959 has zero dtime.  Fix? no
Deleted inode 985960 has zero dtime.  Fix? no
Deleted inode 985961 has zero dtime.  Fix? no
Deleted inode 985962 has zero dtime.  Fix? no
Deleted inode 985986 has zero dtime.  Fix? no
Deleted inode 986013 has zero dtime.  Fix? no
Deleted inode 986029 has zero dtime.  Fix? no
Deleted inode 986031 has zero dtime.  Fix? no
Deleted inode 986032 has zero dtime.  Fix? no
Deleted inode 986035 has zero dtime.  Fix? no
Deleted inode 986041 has zero dtime.  Fix? no
Deleted inode 986046 has zero dtime.  Fix? no
Deleted inode 986048 has zero dtime.  Fix? no
Deleted inode 986050 has zero dtime.  Fix? no
Deleted inode 986053 has zero dtime.  Fix? no
Deleted inode 986054 has zero dtime.  Fix? no
Deleted inode 986057 has zero dtime.  Fix? no
Deleted inode 986069 has zero dtime.  Fix? no
Deleted inode 986070 has zero dtime.  Fix? no
Deleted inode 986071 has zero dtime.  Fix? no
Deleted inode 986078 has zero dtime.  Fix? no
Deleted inode 986092 has zero dtime.  Fix? no
Deleted inode 986109 has zero dtime.  Fix? no
Deleted inode 986130 has zero dtime.  Fix? no
Deleted inode 986182 has zero dtime.  Fix? no
Deleted inode 1147812 has zero dtime.  Fix? no
Deleted inode 1147817 has zero dtime.  Fix? no
Deleted inode 1147843 has zero dtime.  Fix? no
Inode 1147395, i_blocks is 162984, should be 162992.  Fix? no
Deleted inode 3112961 has zero dtime.  Fix? no
Inodes that were part of a corrupted orphan linked list found.  Fix? no
Inode 4194305 was part of the orphaned inode list.  IGNORED.

Pass 2: Checking directory structure
Entry '#sql_15d8_0.MYI' in /entware/tmp (868567) has deleted/unused inode 869517.  Clear? no
Entry '#sql_15d8_0.MYD' in /entware/tmp (868567) has deleted/unused inode 869518.  Clear? no
Entry '#sql_15d8_2.MYI' in /entware/tmp (868567) has deleted/unused inode 869519.  Clear? no
Entry '#sql_15d8_2.MYD' in /entware/tmp (868567) has deleted/unused inode 869520.  Clear? no

Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Block bitmap differences:  -(1785865--1785874) -2000896 -(2304000--2304003) -(2304008--2304015) -(2304040--2304108) -(2304112--2304149) -(2304198--2304229) -(2304422--2304869) -(2304987--2306843) -(2311813--2311837) -2312432 -(2314142--2314239) -(2314256--2314269) -(2315294--2315549) -(2316577--2317310) -(2318079--2318335) -(2318337--2319102) -(2319104--2319345) -(2319351--2320126) -(2322919--2323471) -(2323728--2324479) -(2324484--2326287) -(2327052--2327061) -(2327153--2327154) -(2327156--2327159) -(2327824--2328512) -(2330498--2330623) -(2330927--2330930) -(2330953--2330954) -(2330958--2330963) -(2330965--2330968) -(2330970--2330971) -(2330973--2330975) -(2330977--2330978) -(2330980--2330982) -(2330984--2330987) -(2330989--2330992) -(2330994--2330997) -(2330999--2331027) -(2331029--2331035) -(2331037--2331040) -(2331043--2331045) -(2331047--2331049) -(2331051--2331053) -(2331055--2331057) -(2331059--2331060) -(2331062--2331064) -(2331066--2331068) -(2331070--2331071) -(2331073--2331075) -(2331077--2331079) -(2331081--2331083) -(2331085--2331087) -(2331089--2331090) -(2331092--2331093) -(2331095--2331097) -2331099 -(2331101--2331103) -(2331105--2331107) -(2331109--2331111) -(2331113--2331115) -(2331117--2331118) -(2331120--2331122) -(2331124--2331126) -(2331128--2331130) -(2331132--2331135) -2331137 -(2331394--2332673) -(2333186--2334209) -(2334218--2335234) -(2336258--2336262) -(2338785--2338815) -(2338848--2338850) -(2338857--2338869) -2338882 -(2338884--2338888) -(2338890--2338893) -(2338895--2338897) -(2338920--2338925) -(2338940--2338949) -(2338967--2338973) -(2341028--2342054) -(2345950--2345951) -(2346379--2346442) -(2346699--2347086) -2347176 -(2347180--2347183) -(2463840--2463846) -(2487212--2488319) -(2488327--2488345) -(2488347--2490367) -2496153 -(2574943--2575407) -(2585411--2585415) -(2908138--2908410)
Fix? no

Inode bitmap differences:  -869489 -884806 -983421 -983635 -984792 -984794 -984927 -985956 -(985959--985962) -985986 -986013 -986029 -(986031--986032) -986035 -986041 -986046 -986048 -986050 -(986053--986055) -986057 -(986069--986071) -986078 -986092 -986109 -986130 -986182 -1147812 -1147817 -1147843 -3112961
Fix? no

Free inodes count wrong for group #53 (15087, counted=15086).
Fix? no

optware: ********** WARNING: Filesystem still has errors **********

optware: 79747/7340032 files (2.3% non-contiguous), 1798708/14653670 blocks

should i try
e2fsck -p /dev/sda1
or do i risk to lose some of my webserver files?

 

[Voxel]

as you can see it clearly show my sda1 is mount read-only. and it seems it because of some errors. any idea what can it be ... ssd failure error, cable error, software error?

You have to check the health of you disk. Read-only is mounted if FS has the errors as in your case. Or other choice: SSD usually becomes read-only in the case of pre-failure (automatically). There are some tools to check SSD (as far as I know).

So I'd suggest: First, try to save your files storing them to second disk attached to the router. Second, run e2fsck. And third, check the health of your SSD.

Voxel.

 

[Voxel]

In case anyone out there is:
1) Running an R7800 as an access point
2) About to run an upgrade to Voxel 62
3) Comfortable with telnetting in and running commands

Could you please check the dnsmasq command in top on the AP before and after the upgrade. After the 62 upgrade, my R7800 AP had dnsmasq pinning one of the CPU cores. I shut it down with /etc/init.d/dnsmasq stop. I do not know whether this had been happening on the AP beforehand with Voxel 61.

Dnsmasq on my R7800 router is OK as far as CPU usage. However, the default enabling of the dnsmasq option no-negcache is not appropriate in my situation. There are high numbers of negative reverse lookups from my LAN which flood my DNS provider unless I remove the no-negcache option from /etc/dnsmasq-resolv.conf and reboot.

It is strange because I did not touch in 62SF any dnsmasq settings. I can check it but later. After week or so (my plans are to move one from my R7800 to AP mode).

Voxel.

 

[eliz82]

You have to check the health of you disk. Read-only is mounted if FS has the errors as in your case. Or other choice: SSD usually becomes read-only in the case of pre-failure (automatically). There are some tools to check SSD (as far as I know).

So I'd suggest: First, try to save your files storing them to second disk attached to the router. Second, run e2fsck. And third, check the health of your SSD.

Thanks Voxel. I already copied the files, but it was the database who concern me, because mysql server didn't start because of the (disk) errors, so i could not use "mysqldump" command.
so i copied the "raw database" files from /opt/var/databases

So i tried:
umount /mnt/sda1 - not working because device busy
fuser -km /mnt/sda1 - killed all processed running on sda1
fuser -m /mnt/sda1 - confirm there are no process running on sda1
umount /mnt/sda1 - not working because device busy ... wtf ?

so i was not able to run e2fsck. i had to make another usb optware stick , remove my ssd. then reboot the router, the usb stick was SDA1. i hotplugged SDD and it was mounted as SDB1. Then:

umount /mnt/sdb1 - work ... finally
e2fsck -y /dev/sdb1 - fix all errors without asking
e2fsck -n /dev/sdb1 - checked again there are no error

removed usb stick with SDA1, rebooted the router.
now it seems SSD is mouted as "read-write". but when i accessed my website i had some errors, one of the database had some corrupt tables.

i found this tutorial https://www.a2hosting.com/kb/cpanel...and-repairing-mysql-databases-with-phpmyadmin
i enter phpmyadmin, selected on the tables from my database and repair and optimize.

for now everything seems to be fine. i need to find some tool to check the s.m.a.r.t of my SSD.

 

[eliz82]

i need to find some tool to check the s.m.a.r.t of my SSD.

i found an entware software that can read s.m.a.r.t. , so i could read parameters of my ssd connected to the router.

opkg install smartmontools
smartctl -a /dev/sda1

everything seems to be fine with my ssd even it has 1200 power on days. so it was software error

 

[jsmiddleton4]

Gonna guess you didn't change anything but I'm dropping 2 devices in Attached Device list. Both smart or managed switches. Simple managed switches but managed none-the-less.

They both work fine. Just not showing up. I've booted, power off booted, refreshed DHCP, tried manual assigned IP's, etc.

 

[jsmiddleton4]

I got it working. Still not sure why exactly. Either it was turning back on the IGMP snooping stuff or re-enabling VLAN 1 which I thought was the default already enabled.

 

[Voxel]

everything seems to be fine with my ssd even it has 1200 power on days. so it was software error

BTW I would recommend to use ext4. Journaling is not very good for SSD itself (more write operations) but is much better for safety of your data. See my README re: how to create ext4 for R7800 with e2fsprogs 1.4x

Voxel.

 

[eliz82]

BTW I would recommend to use ext4. Journaling is not very good for SSD itself (more write operations) but is much better for safety of your data. See my README re: how to create ext4 for R7800 with e2fsprogs 1.4x

It remain EXT2 since i got this weird problem with my shell scripts not working. We had a long discussion about this in this topic:
https://www.snbforums.com/threads/c...r-r7800-v-1-0-2-33sf.40198/page-2#post-339024
I first believed is something that has to do with disk. Originally my SSD was ext4, then format to ext3, then format to ext2, then I found that actually the problem is nothing to do with the disk, because of the Notepad++ saving files with Windows end of line (EOL) format not Unix EOL. And I was to lazy to make ext4 again and reinstall all entware software.

 

[Voxel]

I found that actually the problem is nothing to do with the disk, because of the Notepad++ saving files with Windows end of line (EOL) format not Unix EOL. And I was to lazy to make ext4 again and reinstall all entware software.

There is a package in Entware dos2unix:

DOS/Mac to Unix and vice versa text file format converter

Regarding re-install. It is better just to tar your Entware directory on the second USB disk and after re-format restore from tarball.

I am using Entware. Had some problems when my disk was formatted as ext2. After several crashes I started to use ext4. So more than year no any crashes.

Voxel.

 

[percy3]

With .62sf I again have connectivity issues with 2.4 GHz network. With 11k disabled it is better but still from time to time connection is broken, which is surprising since I understand Voxel did not modify any WLAN related packages.
Issue observed with Intel Wireless-AC 7265 card.

 

[vit5421]

Strange. I dont have any 2.4 GHz connectivity issues with Intel 9260 and galaxy s9+. Looks stable as before

 

[jsmiddleton4]

Both channels stable here. IOS clients, PC's.....

 

[kokishin]

@Voxel I bought a Netgear R7800 this weekend. I did a lot of research prior to the purchase and went back and forth on purchasing the Asus RT-AC86U but went with SNB's #1 ranked wireless router. I've been happily using an Asus RT-N66U for years with (the latest) stock firmware. I finally decided it was time to step up from 802.11n to 802.11ac.

Before connecting the R7800 to my WAN, I immediately installed your firmware over the stock Netgear firmware that came loaded on the router. So I have a no real experience with the stock Netgear firmware since I only used it to install your firmware. I got everything set up using your firmware such as dhcp reservations, port forwarding, the 2.4 Ghz and 5 Ghz radios, WAN, nothing too fancy. Everything is working well and seems stable

I've got to admit though that the Netgear firmware that you've improved and stabilized is disappointing compared to Asus's stock firmware. Asus' firmware is much easier to work with especially when setting up dhcp reservations. It seems that the Attached Devices page is not well integrated with the DHCP Reservation page. I had a lot of unidentified devices in the DHCP Reservation page that I had well defined in the Attached Devices page. After all these years, Netgear should be able to do a better job.

I know it's difficult to polish a turd so I want to THANK YOU for all your work to improve the Netgear stock firmware. I've seen a lot of complaints in the Netgear forum about their firmware affecting router stability. Everyone in the know says to use your firmware and I'm glad I did!

 

[percy3]

Looks like problem was affecting only Intel 7265 and only with 2.4 band and is gone for some time already after playing with different driver vendors/versions. Now using Microsoft one (19.51.8.3).

With .62sf I again have connectivity issues with 2.4 GHz network. With 11k disabled it is better but still from time to time connection is broken, which is surprising since I understand Voxel did not modify any WLAN related packages.
Issue observed with Intel Wireless-AC 7265 card.

 

[robwiel]

All seems fine from stock 60 to custom 62, kept my bandwidth ~500/30 with QoS enabled.

On s8+
482/29 (5ghz) but I hope it will be more stable
Right off the bat upload on 2.4ghz is up from 15 to 21.

Will look into dnscrypt and other ad blocking after couple days of testing.
Great job!

Any idea why QoS has this old thing and no new version exist?
Current Performance Optimization Database on Router
Version: v4.1023.gb2747cb
Release Date: October 23, 2017