Photo by Daniel Tuttle on Unsplash
I recently had an issue where AWS sent notifications that they would be stopping EC2 instances on a certain date and time. Turns out, they meant it.
Normally this is no big deal for BOSH deployed VMs. BOSH would notice the VM is missing and recreate it through the CPI with it’s awesome self-healing resurrection feature.
I waited.
Noting happened.
Huh…
Turns out someone had turned off the resurrection feature with the BOSH CLI command:
bosh update-resurrection off
Ah. Ok. To turn resurrection back on I can simply run:
bosh update-resurrection on
But how do I tell what the current state of resurrection is? Is it on? Is it off? BOSH CLI to the rescue? Nope. There is no BOSH CLI command to retrieve the current value.
Where is this kept?
Inside the bosh database of course! There is a table called director_attributes which contains the UUID of the BOSH Director and the current status of resurrection.
This is what it looks like with resurrection disabled:
bosh=# select * from director_attributes;
value | name | id
--------------------------------------+---------------------+----
b0062f10-blah-1224-blah-fd0dcc751234 | uuid | 1
true | resurrection_paused | 2
(2 rows)
A quick bosh update-resurrection on to enable resurrection results in the table looking like:
bosh=# select * from director_attributes;
value | name | id
--------------------------------------+---------------------+----
b0062f10-blah-1224-blah-fd0dcc751234 | uuid | 1
false | resurrection_paused | 2
(2 rows)
Side note: turning resurrection on means the table value is really false. Resurrection off means the table value is really true. Having the field being resurrection_paused caused me to take a double take to make sure I had those values correct.
Second note: if there is no row for resurrection that means that this value has not been set yet. As soon as you run the bosh update-resurrection command the row will be created.