This replaces the use of uci_validate_section() with
uci_load_validate(), which removes the need to declare local variables
for every config option.
This also replaces space indentation with tabs, and removes trailing
whitespace and unnecessary curly brackets.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
fix Makefile chmod (644)
replace MD5SUM with HASH
add PKG_MIRROR_HASH when PKG_SOURCE_PROTO:=git
(PKG_SOURCE_PROTO:=svn tarballs are not reproducible for now)
Signed-off-by: Etienne Champetier <champetier.etienne@gmail.com>
This commit fixes bugs, cleans and enhances init.d script:
1. The first disk and interface was passed to mini_snmpd with extra
leading space.
This bug caused the first monitored disk and interface to be
inaccessible by SNMP daemon.
2. Automatically reload deamon if one of monitored interfaces goes
up/down.
Since mini_snmpd reads interface list only at startup, it won't
detect any interface which appeared after the daemon was started.
Fortunately we can use procd interface triggers to automatically
restart the daemon.
3. Replace hand-written direct ubus calls and json data filtering with
standard network functions.
Signed-off-by: Marcin Jurkowski <marcin1j@gmail.com>
for working out the interface name
Working:
ubus -S call network.interface dump|jsonfilter -e "@.interface[@.interface=\"wan\"].l3_device"
Broken:
ubus -S call network.interface dump|jsonfilter -e "@.interface[@.interface=\"wan\"].device"
Fix run tested:
root@wifi:/overlay/upper# ps |grep mini_snmpd
1404 root 980 S /usr/bin/mini_snmpd -n -c public -L Undisclosed -C VGB <admin@victimsofgaybullying.com> -t 1 -a -d /overlay,/tmp -i br-lan,pppoe-w
Before it wasn't using the pppoe interface it was using the parent
interface eth0 twice. Small 1 line fix. Merge at your convenience.
Signed-off-by: Luke McKee <hojuruku@gmail.com>
Makes use of all available runtime options that can be passed to
mini_snmpd such as -I listen_on interface
Run-tested on trunk on ar71xx. Please review the init script regarding procd
monitoring interfaces, though everything else due to the conversion to
procd should be and done properly, if not overkill compared to most
other script bundled with openwrt.
The main benefit of all this is now larger routers can run multiple
instances of mini_snmpd on different ports, to get around the MIB hard
coded 4 interface/mountpoint limits.
Due to somewhat lacking dev / package maintainer docs except for
http://wiki.prplfoundation.org/wiki/Procd_reference writing this script
took longer than it should. You can see it's evolution here:
https://github.com/hojuruku/openwrt-packages/blob/mini_snmpd/net/mini_snmpd/files/mini_snmpd.init
If everything in it is found to be sane, please include it to the
openwrt wiki on procd as an example.
Due to procd / uci /sbin/validate_data cbi element datatype (uciname) being used
to check uci config, the configuration file now only takes openwrt uci network names not
physical network names as it did before.
http://git.openwrt.org/?p=packages.git;a=commit;h=783e5578ad104d1ca3c31582add08fc8eb4ad083
Like busybox the package Makefile has depends for all runtime
dependencies needed by the init script. mini_snmpd only depends on libc
squashed commits:
mini_snmpd: fix typo to fix procd triggers calling daemon binary directly instead of init script - misuse of $PROG var
mini_snmpd: enable SSP & mini_snmpd to start by default and listen on lan
mini_snmpd: disable RELRO blocker in Makefile after consulation
mini_snmpd: add smarts to init script to detect if ubusd hasn't started yet
mini_snmpd: fix init script - ubus -S doesn't print meaningful errors to give the user
mini_snmpd: Makefile roll back Makefile PKG_RELEASE to 1 for feng shui
Jeffery To
Etienne Champetier
Marcin Jurkowski
Luke McKee