forked from ungleich-public/cdist
35 lines
1.7 KiB
Org Mode
35 lines
1.7 KiB
Org Mode
|
* The scanner, 2020-10-29, Hacking Villa Diesbach
|
||
|
** Motivation
|
||
|
- The purpose of cdist is to ensure systems are in a configured state
|
||
|
- If systems reboot into a clean (think: netboot) state they are
|
||
|
stuck in an unconfigured mode
|
||
|
- We can either trigger *from* those machines
|
||
|
- this is what cdist trigger is for
|
||
|
- Or we can regulary *scan* for machines
|
||
|
- This method does not need any modification to standard OS
|
||
|
** How it works
|
||
|
- cdist scan uses the all nodes multicast group ff02::1
|
||
|
- It sends a ping packet there in regular intervals
|
||
|
- This even works in non-IPv6 networks, as all operating systems
|
||
|
are IPv6 capable and usually IPv6 enabled by default
|
||
|
- Link local is always accessible!
|
||
|
- cdist scan receives an answer from all alive hosts
|
||
|
- These results are stored in ~/.cdist/scan/${hostip}
|
||
|
- We record the last_seen date ~/.cdist/scan/${hostip}/last_seen
|
||
|
- After a host is detected, cdist *can* try to configure it
|
||
|
- It saves the result (+/- logging needs to be defined) in
|
||
|
~/.cdist/scan/${hostip}/{config, install}_result
|
||
|
- If logging is saved: maybe in ~/.cdist/scan/${hostip}/{config, install}_log
|
||
|
- Final naming TBD
|
||
|
** Benefits from the scanning approach
|
||
|
- We know when a host is alive/dead
|
||
|
- We can use standard OS w/o trigger customisation
|
||
|
- Only requirement: we can ssh into it
|
||
|
- Can make use f.i. of Alpine Linux w/ ssh keys feeding in
|
||
|
- We can trigger regular reconfiguration
|
||
|
- If alive && last_config_time > 1d -> reconfigure
|
||
|
- Data can be exported to f.i. prometheus
|
||
|
- Record when configured (successfully)
|
||
|
- Record when seen
|
||
|
- Enables configurations in stateless environments
|