YazFi YazFi v4.x

[abracadabra11]

Still a bit strange with the "unknown" showing up in the client list for 8 of the 10 active 2.4Ghz clients. All YazFi guest clients are assigned IP addresses and names via dnsmasq.postconf. Picks up the name for for two of them the but not others. The unknowns are a mix of Feit smart bulbs and generic band smart plugs. Will see if cycling power to them (at some point) changes anything in the connected guest table.

View attachment 35317

On my side - all my Guest clients have names. But I only have 5 devices connected currently.

 

[Jack Yaz]

Ran in the develop version and all seems fine in a brief look at the CPU. Will monitor. No more regular 50% plus spikes of CPU usage.

View attachment 35315

Edit: Can confirm the sys log being filled with:

Jul 28 16:08:20 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:08:28 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:08:28 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:08:36 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:08:36 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:08:44 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:08:44 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:08:52 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:08:52 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:09:00 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:09:00 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:09:07 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:09:07 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)

Still a bit strange with the "unknown" showing up in the client list for 8 of the 10 active 2.4Ghz clients. All YazFi guest clients are assigned IP addresses and names via dnsmasq.postconf. Picks up the name for for two of them the but not others. The unknowns are a mix of Feit smart bulbs and generic band smart plugs. Will see if cycling power to them (at some point) changes anything in the connected guest table.

View attachment 35317

good to hear cpu spikes have gone. I'm looking into why your IP addresses are coming up as unknown from the samples you DM'd me

 

[bennor]

Here's the behavior I'm seeing:
2. Navigating to the YazFi GUI starts the spiking that was exhibited in 4.3.1. Cores 1 and 2 generally spike to around 40% each when on the page. System log shows recurring YazFiconnected clients service start and end every 5 seconds.
3. Deselecting 'Update Automatically' under Connected Guests menu eliminates the spiking and stops the recurring YazFiconnected clients service.

Will echo these observations. Seeing the same.

CPU usage spikes when YazFiGUI page loaded (and Table will refresh every 5s is checked).

Unchecking Table will refresh every 5s, CPU usage drops. Drop off in chart corresponds to unchecking table refresh.

Unchecking the table refresh also stops the sys log entries. Note the four minute gap between 16:20:59 and 16:24:18 where I had unchecked the table refresh. And leaving the YazFi GUI page also stops the sys log entries (it appears).

Jul 28 16:20:34 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:34 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:20:42 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:42 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:20:50 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:50 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:20:59 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:59 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:24:18 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:24:18 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:24:25 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:24:25 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:24:33 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:24:33 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)

Unchecking the table refresh check box gets rechecked when leaving the YazFi GUI page and then returning to it or when reloading the page. (see gap from 16:31:43 to 16:34:10)

Jul 28 16:31:35 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:31:43 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:31:43 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:34:10 dnsmasq-dhcp[17859]: DHCPREQUEST(br0) 192.168.2.13 cc:40:d0:3a:f4:37
Jul 28 16:34:10 dnsmasq-dhcp[17859]: DHCPACK(br0) 192.168.2.13 cc:40:d0:3a:f4:37 NetgearGS908E
Jul 28 16:34:10 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:34:10 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:34:14 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:34:14 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)

 

[Jack Yaz]

Will echo these observations. Seeing the same.

CPU usage spikes when YazFiGUI page loaded (and Table will refresh every 5s is checked).

View attachment 35318

Unchecking Table will refresh every 5s, CPU usage drops. Drop off in chart corresponds to unchecking table refresh.

View attachment 35319

Unchecking the table refresh also stops the sys log entries. Note the four minute gap between 16:20:59 and 16:24:18 where I had unchecked the table refresh. And leaving the YazFi GUI page also stops the sys log entries (it appears).

Jul 28 16:20:34 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:34 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:20:42 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:42 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:20:50 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:50 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:20:59 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:20:59 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:24:18 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:24:18 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:24:25 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:24:25 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:24:33 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:24:33 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)

Unchecking the table refresh check box gets rechecked when leaving the YazFi GUI page and then returning to it or when reloading the page. (see gap from 16:31:43 to 16:34:10)

Jul 28 16:31:35 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:31:43 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:31:43 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:34:10 dnsmasq-dhcp[17859]: DHCPREQUEST(br0) 192.168.2.13 cc:40:d0:3a:f4:37
Jul 28 16:34:10 dnsmasq-dhcp[17859]: DHCPACK(br0) 192.168.2.13 cc:40:d0:3a:f4:37 NetgearGS908E
Jul 28 16:34:10 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:34:10 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)
Jul 28 16:34:14 rc_service: httpd 25424:notify_rc start_YazFiconnectedclients
Jul 28 16:34:14 custom_script: Running /jffs/scripts/service-event (args: start YazFiconnectedclients)

this is expected, i did say syslog would get spammed while on the page

 

[abracadabra11]

its worth mentioning despite the CPU usage, the process priority is bumped to 15 niceness so when your VPN demands go up, it should push the YazFi process to the back of the queue

Did a quick test similar to the test I did on 4.3.1 and I'm actually getting worse performance on the develop branch. VPN throughput is ~15-18% slower when on the YazFi GUI page with the clients refreshing. VPN throughput is nominal when off the page or clients refresh is deselected.

This doesn't matter for my purposes, but providing the feedback for reference.

 

[Jack Yaz]

Did a quick test similar to the test I did on 4.3.1 and I'm actually getting worse performance on the develop branch. VPN throughput is ~15-18% slower when on the YazFi GUI page with the clients refreshing. VPN throughput is nominal when off the page or clients refresh is deselected.

This doesn't matter for my purposes, but providing the feedback for reference.

huh, weird. maybe the service calls are adding to the loading. if you watch top/htop you should see the YazFi process marked N 15

 

[abracadabra11]

huh, weird. maybe the service calls are adding to the loading. if you watch top/htop you should see the YazFi process marked N 15

I don't know enough about how top works, but the data I'm seeing is pretty baffling.

YazFi doesn't seem to consume even 1% of CPU according to top (even when on the YazFi GUI page with clients refreshing), which is very different from the utilization displayed on the router GUI. Speed tests using OpenVPN only utilize about 25% of CPU according to top as well, which also doesn't make much sense since that process is clearly CPU-limited.

I didn't see N15, only noted an N under the STAT column (but it looks like any additional entries after N are deprecated on the screen).

 

[Jack Yaz]

I don't know enough about how top works, but the data I'm seeing is pretty baffling.

YazFi doesn't seem to consume even 1% of CPU according to top (even when on the YazFi GUI page with clients refreshing), which is very different from the utilization displayed on the router GUI. Speed tests using OpenVPN only utilize about 25% of CPU according to top as well, which also doesn't make much sense since that process is clearly CPU-limited.

I didn't see N15, only noted an N under the STAT column (but it looks like any additional entries after N are deprecated on the screen).

what % values are listed across the top for io, use, sys etc.?

opkg install htop

if you have entware installed htop is generally a nicer tool

 

[bennor]

As another user data point. Yeah not seeing "N15" either when running top. Not running Entware (I don't think) either. If I leave top running and have the YazFi GUI page up I see occasional YazFi service_event entries show up as the second, third, or fourth entry. Examples:

  PID  PPID USER     STAT   VSZ %VSZ CPU %CPU COMMAND
   27     2 john    SW       0  0.0   1  3.7 [mtdblock3]
  240     1 john    R     6772  2.6   0  2.3 nt_center
  216     1 john    S      812  0.3   1  2.2 /usr/sbin/jitterentropy-rngd -p /var/run/jitterentropy-rngd.pid
  743     1 john    S     7644  2.9   1  1.3 watchdog
8680     1 john    S N   1732  0.6   1  1.3 {YazFi} /bin/sh /jffs/scripts/YazFi service_event start YazFiconnectedclients
25424     1 john    S     7068  2.7   1  0.2 httpd -i br0
7302 19840 john    R     1452  0.5   1  0.2 top

  PID  PPID USER     STAT   VSZ %VSZ CPU %CPU COMMAND
   27     2 john    SW      0  0.0   1  3.4 [mtdblock3]
  216     1 john    S      812  0.3   1  2.1 /usr/sbin/jitterentropy-rngd -p /var/run/jitterentropy-rngd.pid
14808     1 john    S N   1732  0.6   0  1.8 {YazFi} /bin/sh /jffs/scripts/YazFi service_event start YazFiconnectedclients
  240     1 john    S     6772  2.6   1  1.7 nt_center

Some of these issues don't really affect or bother me, others may be a performance issue depending on one's configuration.

 

[abracadabra11]

opkg install htop

if you have entware installed htop is generally a nicer tool

Installed htop - much nicer tool indeed.

I'm generally seeing YazFiconnectedclients service start utilizing ~-2-5% CPU according to htop CPU% column. But the processors listed up top show a much larger utilization (in line with Asus GUI) of around 25-70% of Cores 1 and 2.

 

[HappyJuicy]

Hi,

Does two way access is fixed? Also the settings in VPN director not persistant after a reboot?

 

[Jack Yaz]

Hi,

Does two way access is fixed? Also the settings in VPN director not persistant after a reboot?

Two way isn't broken as far as I know so you'll need to provide more information.

Make sure you've updated YazFi with any hotfixes but again, not aware of it not being persistent

 

[bennor]

Hi,

Does two way access is fixed? Also the settings in VPN director not persistant after a reboot?

To echo what Jack wrote. What issue or problem are you, or were you, having with Two Way Access in YazFi.

VPN Director is a feature of Asus-Merlin firmware not the YazFi add-on script. If you are having a specific issue or problem with VPN Director that is unrelated to your YazFi usage you really should inquire about your issue with VPN Director in the main Asus-Merlin subforum. Or search through, if you haven't done so already, the VPN Director discussions in the Asus-Merlin subforum to see if your issue has been previously discussed.

https://www.snbforums.com/forums/asuswrt-merlin.42/

 

[bennor]

Bit of a chuckle this morning. I assume some updating is going on with YazFi.
Updated through the YazFi CLI to: "hotfix available - v4.3.2"
For giggles lets run a check via the YazFi GUI after updating to that hotfix and see what it indicates...

OK so lets go back to the CLI and check for an update again. This time it indicates...
"New version of YazFi available - v4.3.1"

 

[Jack Yaz]

Bit of a chuckle this morning. I assume some updating is going on with YazFi.
Updated through the YazFi CLI to: "hotfix available - v4.3.2"
For giggles lets run a check via the YazFi GUI after updating to that hotfix and see what it indicates...
View attachment 35343

OK so lets go back to the CLI and check for an update again. This time it indicates...
"New version of YazFi available - v4.3.1"

im getting ready to merge develop to master while on my lunch break, so develop users if they are over zealous on the update button will see some oddities until I've had time to merge (I went to make a sandwich, sue me!)

 

[Jack Yaz]

v4.3.2 is now available
Changelog:

• CHANGED: Connected client information will now only update when looking at the WebUI page for YazFi with auto refresh enabled

 

[roscoe211]

I was seeing very high CPU usage after upgrading. Reboot and let it sit overnight, still seeing higher than normal from:


255 1 user S 812 0.3 0 14.5 /usr/sbin/jitterentropy-rngd -p /var/run/jitterentropy-rngd.pid


Load average is about 1.5 when everything is idle, when it is typically ~.1 or less on my 68u.

 

[bennor]

I was seeing very high CPU usage after upgrading. Reboot and let it sit overnight, still seeing higher than normal from:

If you haven't updated to v4.3.2 that was released today, do so. When the YazFi GUI page is loaded (and or when the YazFi client information module is running) there will be periodic spikes in CPU usage (something discussed over the last three pages of this thread) but those spikes should go away, with the latest YazFi version. One should now only see the CPU pike usage when one is on the YazFi GUI page and stop when one leaves that page or one unchecks "Table will refresh every 5s" while on the YazFi GUI page.

 

[roscoe211]

Updated to 4.3.2 and we're back to normal. Thank you!

 

[Jack Yaz]

v4.3.3 is now available
Changelog:

• NEW: Generate QR Code for guest network at the command line (requires Entware for qrencode package) - credit @RMerlin for QR code function in firmware WebUI for inspiration