Null output in nagios plugin when warning or critical

[unix196]

We have nagios3 in docker (~3400 checks). Added vault monitoring in nagios:

...
define service {
    use                     http-service
    host_name               some_server
    service_description     vault: expire token accessor token-for-nagios
    servicegroups           vault
    check_command           check_vault!token-lookup -address=some_server -token=s.O...5CK -token-accessor=ZQ...TA -output=nagios -warning=768h -critical=72h
}

cat /etc/nagios-plugins/config/vault.cfg
define command {
   command_name     check_vault
   command_line     /usr/lib/nagios/plugins/hashicorp-vault-monitor $ARG1$ 
}

When I try run check in CLI, everything is working:

 /usr/lib/nagios/plugins/hashicorp-vault-monitor token-lookup -address=some_server -token=s.O...5CK -token-accessor=ZQ...TA -output=nagios -warning=768h -critical=72h
vault WARNING - This (renewable) token will expire on Sat, 09 May 2020 05:32:23 UTC (1 week 5 days 1 hour 19 minutes 36 seconds left)
echo $?
1

Some other token:

/usr/lib/nagios/plugins/hashicorp-vault-monitor token-lookup -address=some_server -token="s.O...K" -token-accessor="hd...8" -output=nagios -warning=168h -critical=72h
vault OK - This (renewable) token will expire on Mon, 21 Jan 2030 08:52:06 UTC (9 years 38 weeks 5 days 4 hours 23 minutes 13 seconds left)
echo $?
0

In nagios web interface I see that - https://ibb.co/1K76F7p. I try in my check "vault: expire token accessor token-for-nagios" add single and double quotes, but problem with null output is remain. When check is generate null in web interface, I see her output in docker logs:

docker logs --tail 20 -f nagios3
vault WARNING - This (renewable) token will expire on Sat, 09 May 2020 05:32:23 UTC (1 week 5 days 49 minutes 28 seconds left)

I don't think that problem with nagios in docker ( we now run 3000 different checks without any trouble) + when hashicorp-vault-monitor generate "OK" - I see that in web interface. Trouble only when plugin generate "Warning" or "Critical" (also null outputs will be in other checks - I check that).

usr/lib/nagios/plugins/hashicorp-vault-monitor --version
HashiCorp Vault Monitor v0.8.4 ('7b2326ea73281891139e077aa39f2d91f83c493c+CHANGES')

For example, I found thread - https://www.linuxquestions.org/questions/linux-software-2/nagios-interprets-perl-plugin-output-as-null-948605/, where similar problem (plugin written on perl)

My issue is resolved. My plugin does file IO and wasn't opening a file for reading. Works from the command line possibly because I ran the script from the same directory as the script. Nagios runs the script from absolute path in another working directory.

May be same problem (or similar) in that plugin?

[madrisan]

Thank you for the report. I've an idea of the root cause of this issue...

[madrisan]

Is it possible to add 2>&1 at the end of nagios command_line definition? Does this fix the null output issue?

[unix196]

Yes, this working! That's basically enough, but better mention in documentation.

[madrisan]

I will fix this directly in the code. This is just a workaround and a test to see if I got the cause of the problem.