Sense client error "previous check command execution in progress"

Question

I am using latest sensu core version 0.26.1, set up sensu server on one CentOS machine, set up one sensu client on another.

There are about 500 checks in one sensu client, I keep seeing "previous check command execution in progress" in sensu-client.log, but each check actually finishes very fast (most of them less than 0.1 seconds and interval is 60 seconds) , i can confirm this by running it under sensu user sudo su sensu -c "{run my check}".

However, the actual situation was, in the uchiwa dashboard, it shows many of my checks were running more than 1 minute, seems the sensu client got seriously stuck/slow, i tried the following approaches:

1. Remove several long running checks and restart sensu server/client
2. Add timeout definition to my checks, limit timeout to 10, then resulted in many of the checks "Execution timed out".
3. I tried only run 1, 10, 50 checks, seems everything was working normally, however, as soon as the count of the checks reaches certain number maybe 200-300, the issue occurred.

Neither of above work, is there a way to debug which check(s) are actually blocking? Or can I configure sensu to simply kill the check when it exceeded the timeout definition? So that I won't see this message in log "previous check command execution in progress"

I got blocked by this and need help:)

Cumshaw

Answer 1

To debug this you can obviously check the sensu-client.log. You can set up the log level in the defaults file for sensu

besides that: ps aux | grep sensu will show you that every check will show up as a separate process and you can easily manually confirm if and which checks maybe hung up. The usual problem here is rather that there are flaws in the checks which cause them to hang than the sensu-client itself hanging up.

I have experienced that myself with some checks.

To get more in depth with your issue at least some output of the logs and the ps command [ when the issue occurs ] will be needed.

Marable