check_mk_agent generates heartbeat errors

[Jeffrey Lewis]

After installing heartbeat + pacemaker, running /usr/bin/check_mk_agent generates the following errors in /var/log/syslog,

Jul 25 16:54:43 jlewis1 heartbeat: [11096]: ERROR: api_process_request: bad request [getrsc]
Jul 25 16:54:43 jlewis1 heartbeat: [11096]: ERROR: MSG: Dumping message with 5 fields
Jul 25 16:54:43 jlewis1 heartbeat: [11096]: ERROR: MSG[0] : [t=hbapi-req]
Jul 25 16:54:43 jlewis1 heartbeat: [11096]: ERROR: MSG[1] : [reqtype=getrsc]
Jul 25 16:54:43 jlewis1 heartbeat: [11096]: ERROR: MSG[2] : [dest=jlewis1]
Jul 25 16:54:43 jlewis1 heartbeat: [11096]: ERROR: MSG[3] : [pid=11533]
Jul 25 16:54:43 jlewis1 heartbeat: [11096]: ERROR: MSG[4] : [from_id=11533]
Jul 25 16:54:43 jlewis1 cl_status: [11533]: ERROR: Cannot get cluster resource status
Jul 25 16:54:43 jlewis1 cl_status: [11533]: ERROR: REASON: Resource is managed by crm.Use crm tool to query resource

I expected to be able to run /usr/bin/check_mk_agent without a slew of errors showing up in /var/log/syslog related to heartbeat.

STEPS TO REPRODUCE:
1- launch Ubuntu 12.04 LTS node
2- aptitude install heartbeat
3- /bin/dd if=/dev/urandom bs=512 count=1 status=noxfer 2> /dev/null | /usr/bin/openssl dgst -md5 | /usr/bin/awk '{print "auth 1\n1 sha1",$2}' > /etc/ha.d/authkeys ; /bin/chmod 0600 /etc/ha.d/authkeys

4- vim /etc/ha.d/ha.cf
autojoin none
coredumps false
keepalive 2
deadtime 15
initdead 60
warntime 5
pacemaker respawn
node example1
ucast eth0 10.11.12.13

5- service heartbeat start

6- sleep 60 ; crm status

7- aptitude install check-mk-agent

8- tail -f /var/log/syslog

9- /usr/bin/check_mk_agent

Other Data:

root@jlewis1:/etc/ha.d# lsb_release -rd
Description: Ubuntu 12.04.2 LTS
Release: 12.04

root@jlewis1:/etc/ha.d# apt-cache policy check-mk-agent
check-mk-agent:
  Installed: 1.1.12-1ubuntu1
  Candidate: 1.1.12-1ubuntu1
  Version table:
 *** 1.1.12-1ubuntu1 0
        500 http://mirrors.service.networklayer.com/ubuntu/ precise/universe amd64 Packages
        100 /var/lib/dpkg/status

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Robert Sander (gurubert)]

This happens because the check_mk agent looks for crm and cl_status CLI tools.

crm is in package pacemaker and cl_status in heartbeat.

If you have setup your cluster with heartbeat+pacemaker you have both CLI tools, but only crm will work.