Syslog not always working after a reboot

[Jack Yaz]

Sometimes SysLog will fail to generate logs on a reboot. If I hit Apply on the settings, it starts working and shows:

Aug 24 22:51:09 kernel: klogd started: BusyBox v1.25.1 (2017-08-18 17:40:15 EDT)
Aug 24 22:51:10 rc_service: httpd 494:notify_rc restart_logger
Aug 24 22:51:10 kernel: klogd: exiting
Aug 24 22:51:10 kernel: klogd started: BusyBox v1.25.1 (2017-08-18 17:40:15 EDT)

Where can I start troubleshooting to find why klogd isn't always starting, if that is the problem?

 

[Jack Yaz]

I definitely seem to have an issue with klogd not starting, I'm guessing that's not a good thing

 

[Jack Yaz]

Factory restore and nvram restore later, now loading. However a new issue, its only showing the last 5-6 lines at a time, unless I delete the syslog files and restart klogd...any ideas?

 

[ColinTaylor]

Running out of RAM (/tmp) or JFFS? Try a df -h

 

[john9527]

klogd is the kernel logger.....most of the logging is done by syslogd. Are you sure you didn't accidentally change the loglevel (Administration > System)?

 

[Jack Yaz]

Running out of RAM (/tmp) or JFFS? Try a df -h

Over 100mb ram available and jffs is using about 10mb out of 62

klogd is the kernel logger.....most of the logging is done by syslogd. Are you sure you didn't accidentally change the loglevel (Administration > System)?

Log settings are set to
Default Notice
Log more urgent than Debug

 

[Jack Yaz]

Aug 25 06:34:32 kernel: usbcore: deregistering interface driver rndis_host
Aug 25 06:34:32 kernel: usbcore: deregistering interface driver cdc_ether
Aug 25 06:34:32 kernel: usbcore: deregistering interface driver asix
Aug 25 06:34:32 kernel: cdrom: Uniform CD-ROM driver unloaded
Aug 25 06:34:32 kernel: usbcore: deregistering interface driver usblp
Aug 25 06:34:32 kernel: xhci_hcd 0000:00:0c.0: remove, state 1
Aug 25 06:34:32 kernel: usb usb1: USB disconnect, address 1
Aug 25 06:34:32 kernel: xhci_hcd 0000:00:0c.0: USB bus 1 deregistered

Thats the last entries before a reboot via GUI. Then logging just doesn't start

 

[john9527]

There's basically only two thing that happen before the loggers are started on a reboot....
jffs is initialized
the init-start script is run

My first guess would be the second.....check (or disable) your init-start script if you have one to see if it has an error.

 

[Jack Yaz]

I've established a factory reset lets logging work. A clean restore using nvram 25a, reboot, and no more logging. I've disabled all startup scripts in JFFS other than Skynet and FreshJR QoS, still no logging.

This sounds like there's something wrong with either my nvram variables, or something else in JFFS. Is it safe to delete the syslog.log files? What permissions should they have? I wonder if it's not being set correctly

 

[Jack Yaz]

There's basically only two thing that happen before the loggers are started on a reboot....
jffs is initialized
the init-start script is run

My first guess would be the second.....check (or disable) your init-start script if you have one to see if it has an error.

init-start contains just these lines, and has done for several versions without issue :/

# Decrease swappiness for emergency buffer only
echo 10 > /proc/sys/vm/swappiness

nvram set clkfreq=1200,800
nvram commit

 

[john9527]

try taking out the clkfreq and commit....Asus has continually been locking that down.

 

[Jack Yaz]

try taking out the clkfreq and commit....Asus has continually been locking that down.

I think you have cracked it! I've left it services-stop (I saw in a thread somewhere to put it there), and while the GUI doesn't show it, BogoMIPS does...so I'm guessing init-start was failing due to some kind of new lockdown!

 

[john9527]

Really double check the BogoMIPS.....last folks I worked with, any attempt to change the clkfreq (even in services-stop) actually resulted in the router underclocking

 

[Jack Yaz]

It's showing just under 2400, so 1200 x 2

Is that right

 

[john9527]

Nope....that's underclocked.....should be close to 2000 per core (usually 1997 or 1998). You may need to do another factory reset to get it back to normal (after taking it out from services-stop).

 

[john9527]

On my fork, I can still overclock.....here's the numbers for 1200,800
2400 per core. Use the cat command

cat /proc/cpuinfo
Processor       : ARMv7 Processor rev 0 (v7l)
processor       : 0
BogoMIPS        : 2398.61

processor       : 1
BogoMIPS        : 2398.61

EDIT: Stock clock 1000,800 is the approx 2000 bogomips

 

[Jack Yaz]

On my fork, I can still overclock.....here's the numbers for 1200,800
2400 per core. Use the cat command

cat /proc/cpuinfo
Processor       : ARMv7 Processor rev 0 (v7l)
processor       : 0
BogoMIPS        : 2398.61

processor       : 1
BogoMIPS        : 2398.61

ASUSWRT-Merlin RT-AC87U 380.68-0 Fri Aug 18 21:40:17 UTC 2017
admin@RT-AC87U:/tmp/home/root# cat /proc/cpuinfo
Processor       : ARMv7 Processor rev 0 (v7l)
processor       : 0
BogoMIPS        : 2398.61

processor       : 1
BogoMIPS        : 2398.61

Features        : swp half thumb fastmult edsp
CPU implementer : 0x41
CPU architecture: 7
CPU variant     : 0x3
CPU part        : 0xc09
CPU revision    : 0

Hardware        : Northstar Prototype
Revision        : 0000
Serial          : 0000000000000000
admin@RT-AC87U:/tmp/home/root#

 

[Jack Yaz]

This is what mine shows with 1200,800 in services-stop only

 

[john9527]

That looks good....I'm a bit surprised that the services-stop is working for you, but as the saying goes, don't look a gift horse in the mouth

 

[Jack Yaz]

Haha definitely! Tbh if I wanted serious performance I'd build myself a pfSense box. The RT-AC87U will do for my current uses though!