Why did Lassie reboot my Linode?
My Linode was rebooted, we didn't initiate a reboot.
The error message said:
Lassie initiated boot: ExampleLinode Entered: 2 hours 14 minutes ago - Took: 16 seconds
Lassie will never initiate a shutdown or reboot of your Linode. It only will ever bring your server back up in the case that it is unexpectedly down.
For more context, a Lassie initiated boot only occurs if Lassie detects that your Linode is already powered off and there was no shutdown command issued from the Linode Manager or API. So if you did not shutdown your Linode in any way, something must have gone wrong with your Linode resulting in an unexpected shutdown.
There aren't really any logs because by the time Lassie detects that something is wrong with the server it is already powered off. Due to this there are no logs to collect besides the timestamp of when Lassie detected and then rebooted your Linode.
To see why this shutdown happened initially, you can check in the normal places from within your Linode:
Most helpful will probably be:
Or if you suspect some type of kernel panic:
You can also check specifically for logs for any services that were running that could have caused an issue. A webserver like nginx for example:
The following is modified from our Blog Post
October 26, 2007 11:18 am
Lassie, the shutdown watchdog
Lassie – Linode Autonomous System Shutdown Intelligent rEbooter
Lassie is a shutdown watchdog. It will automatically reboot your Linode if it powers off unexpectedly. It works by detecting that your Linode has powered off without a shutdown job being responsible for the power off condition. To prevent a loop, it will give up after 5 boots have occurred within a 15 minute period.
It’s enabled by default, but the setting can be changed on your Linode's Settings page[/url]
A side effect of this is that now “reboot” from inside your Linode will function as expected, but a “shutdown” from inside your Linode will result in Lassie kicking in and issuing a boot. Previously both of those commands resulted in a shutdown. I can’t win :).
A work-around for this is to issue a shutdown job via Lish, by creating an alias for “shutdown” to do: “ssh LinodeUsername@hostXX.linode.com shutdown”. You can find the name of the host your Linode runs on by looking at box labeled "Host" in the lower right of your Linode's dashboard.
I have also received unexpected shutdown/reboot by Lassie.
Note that the above says "Lassie will never initiate a shutdown or reboot of your Linode." - but I'm not sure that's 100% true.
Another post says "When … communication is disrupted in certain ways indicative of an unresponsive host, Lassie will … reboot … that Linode"
There's nothing in any of my logs suggesting errors, so either the hypervisor killed it randomly; or my kernel panicked such that it could not write anything to the log; or the IO failed; …
It's not happened lots so I'll keep an eye on it.
@rich I'm sorry to hear that you have previously experienced unexpected reboots due to Lassie's watchdog capabilities.
Both posts are correct, and I am happy to clear up any misunderstanding. Lassie will never cause your Linode to shut down. Lassie does restart a Linode that has not been shut down properly. Sometimes, this can also happen when Lassie doesn't receive a response from the Linode itself which can be due to a few factors.
Next time this occurs, I recommend running a file system check in order to see if there are any problems with the Linode.
Hope this clears things up for you!
Could you please elaborate about the other factors that can trigger lassie by the false positive? We also had two lassie initiated restarts in 2 days. There nothing suspicious in the logs, no file system problems whatsoever.
Howdy all! I've come across some information I hope will help you locate what happened inside your machine that would trigger our Linode Autonomous System Shutdown Intelligent rEbooter, (Lassie for short).
If your system is running systemd, you can use the journalctl service.
journalctl | grep -i shutdown, and that will bring up any shutdown related activities. Here are two examples I've seen:
Mar 03 08:24:02 TestNode sudo: testUser : TTY=pts/0 ; PWD=/home/testUser ; USER=root ; COMMAND=/sbin/shutdown -h now
The above line shows that a user named "testUser" used sudo and issued the
shutdown -h now command which brings the computer to a halt and then, Lassie to the rescue!
Feb 26 10:37:14 TestNode2 systemd: Starting LXD - container startup/shutdown...
Feb 26 10:37:14 TestNode2 systemd: Started LXD - container startup/shutdown.
This shows that the system user "systemd" did something with LXD, which caused the Linode to shutdown. While we can't yet confirm what caused LXD to issue this job, we can confirm on the host level that the hypervisor detected a shutdown (not issued by the API) on the Linode which triggers Lassie.
If anyone comes across any other examples please be sure to list them here!