I have created a simple sensor and automation to detect if there are any login failures on the front end. This is a proof of concept to increase the security awareness of HASS. This is in responce to my feature request posted to : IP Security Logs IP Security Logs
I have this triggering a notification though telegram as an example. This is only proof of concept as more information could be extracted from the logs to include in the sensor if one was so inclined.
I get the files that are in that directory, including home-assistant.log
I tried just running the command as user PI and Hass and both return a value of 1. (1 should be the correct value as the error from the log I pasted above contains âinvalidpasswordâ)
Just a small precision , the above wonât work when the logger is set to Info as the whole command is logged in home-assitant.log so it will detect the trigger action logging and then count it in for a bad password attempt
This popped up again and It thought after a fresh install of jessie and upgrading to 31.1 this should work. No dice.
homeassistant.components.sensor.command_line: Command failed: grep -c "Login attempt or request with an invalid password" /home/hass/.homeassistant/home-assistant.log
I tried philippe_audetâs version too:
platform: command_line
command: grep -c "Login attempt or request with an invalid password" /home/hass/.homeassistant/home-assistant.log
name: badlogin
Anyone else getting an error with this, or am I the only one?
You are right , it was a fake good condition.
After much back and forth with the config and such i got it working
First off , the message you get : [quote=âsilvrr, post:10, topic:4280â]
homeassistant.components.sensor.command_line: Command failed: grep -c âLogin attempt or request with an invalid passwordâ /home/hass/.homeassistant/home-assistant.log
[/quote]
does not mean much as itâs the same message either your command failed, you donât have the permission , etc âŚetc âŚ
Another thing , i get the command failed message as well when i try to grep the /var/log/syslog file for which hass have the access.
So all in all any error will give this message
Now for the solution to work
To be noted : you will need to pollute your logs which for me is a draw back
So the solution for it to work with the command line is to fetch for a string without any spaces within the home-assistant.log file.
to be able to succeed without getting your own grp command in the count of bad login failure hereâs what you need to do
logger:
default: critical
logs:
homeassistant.core: info
with the above you will be able to get the error in your count
but again with that level of login the home assistant log file will become large pretty fast
Would be better if the command_line plateform would let us do the same in the /var/log/syslog file since everything is logged in there ⌠seems to be a limitation as the same command as above fails when trying to fetch the info from there even with all the proper perms,etc âŚ
I just try with a symlink just in case ⌠and it worked after i changed the permissions to put read to all ( as sudo is not supported )
So i set back the HA logger to critical and now the only thing left is that i need to grep first from the last reboot and then grep for the error in the log
Finally got it working , donât ask me why though
I was about to open a bug to HS, i was repeating the steps by steps to give as much information as possible and âbamâ it worked
First off i updated to 0.32.3 this morning
second, the first entry will always be a failed command message since the grep command did not find any entry ( this could be reported as a bug since normally it should just report a 0 as on the command line
Third , the looked for string is not valid anymore as the error message changed from a few version down the road
Result at startup 16-11-11 14:15:59 homeassistant.components.sensor.command_line: Command failed: grep -c 'Login attempt' /home/hass/.homeassistant/home-assistant.log
Result when login failed
16-11-11 14:17:07 homeassistant.components.http: Login attempt or request with an invalid password from 70.83.11.62
16-11-11 14:17:08 homeassistant.components.http: Login attempt or request with an invalid password from 70.83.11.62
16-11-11 14:17:11 homeassistant.components.http: Login attempt or request with an invalid password from 70.83.11.62
Screen shot of the result in HA
As you can see, the result will always be +1 as it the actual bad login attempt
Also , if you reduce the Logger level ( or the component level ) above âwarningâ the failed command will still be logged in but not the entry as above, so you will always staty with Value = 1 with your grep -c
Hopefully it will help some of you
Also, i will submit an updated documentation as the example on the site is not up to date anymore
The following loggger setting will remove the +1 caused by the first command which is failing and will give you the proper count
if you can live with those http log setting
Any way to do some kind of fail2ban setup? Too many wrong passwords and the IP gets locked out for a certain amount of time? Might be useful to prevent brute-force attempts as the project continues to grow
Thats a good solution. I may implement this, but instead of stopping HA just shut down my Raspberry Pi, that way the device they are trying to access is gone along with the HA instance.
Anyone know how to attach a file (home-assistant.log) to a STMP notification. It would be nice to get the log file so I can see what the IP address is.
@silvrr
If you get to have this running please share your findings
I have not been able to issue a sudo reboot command from HA and I did not want to grant my hass user root access to perform it
I noticed after my update to 34.3/4 that there is a persistent notification for a failed login attempt. The state of this trigger includes the IP the failed login was made from.
Anyone have any idea on how to automate this for a notification? Only part I donât have is the trigger. Not sure how to trigger off a persistent notification.
You could try a state trigger with entity_id: persistent_notification.httplogin. Donât specify state or from/to on this trigger as the content of the persistent notification is variable. So something like this:
Bare in mind I havenât tested the code above, so it may contain errors. The template condition is there to avoid sending a notification when dismissing the persistent notification (at which point the state of the persistent notification goes from Invalid password used from xxx.xxx.xxx.xxx to None).
The alternative for this condition would be to include from: 'None' in the trigger, but then you risk not getting notified if the persistent notification is overwritten before it gets dismissed.
And, of course, replace the pushover notify service in my code with whatever notify service you use.