Re: [B.A.T.M.A.N.] alfred dies without any log/dmesg entry

[Sven Eckelmann <sven@narfation.org>]

[-- Attachment #1: Type: text/plain, Size: 1526 bytes --]

On Wednesday 30 December 2015 13:32:39 3zl Trizonelabs wrote:
> ref: alfred ver 2015.1 / .2 @openwrt CC
> 
> afred.facters disabled
> 
> alfred process dies (not listed in ps ) without leaving any log entry
> 
> does anybody have any hint or idea ?

I don't use the new OpenWrt init scripts and therefore hoped that one of the 
authors would answer here.

But maybe it is an alfred crash. You can try to find more about the reasons of 
a crash by enabling core dumps [1]. I think the netifd init script should then 
enable the rest ("ulimit -c unlimited" and setting 
/proc/sys/kernel/core_pattern).

You should then check /tmp for *.core files when you've detected that alfred 
isn't running anymore. If you find some from the alfred process then you can 
assume that there was some crash. You can analyze these core dumps as 
described in the OpenWrt wiki [2]. Just specify the core file instead of 
"192.168.1.1:9000" as the first parameter for ./scripts/remote-gdb. 

You may also want to change the alfred/Makefile before compiling your image to 
make debugging easier. Try replacing the lines

    TARGET_CFLAGS  += -ffunction-sections -fdata-sections -flto
    TARGET_LDFLAGS += -Wl,--gc-sections -fuse-linker-plugin

With

    TARGET_CFLAGS += -ggdb3 -O0

to make the output of things like "bt full" in remote-gdb a lot better to 
read.

Kind regards,
	Sven

[1] OpenWrt config: CONFIG_KERNEL_ELF_CORE=y
    under "Global build settings" as "Enable process core dump support"
[2] https://wiki.openwrt.org/doc/devel/gdb

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 819 bytes --]