crm does not monitor haproxy

[Michał Ajduk]

The way that hacluster resources are organized allows haproxy to be down on a node whre the VIP is running.

Example configuration:
root@juju-3f2dbe-21-lxd-7:~# crm config show
node 1000: juju-3f2dbe-23-lxd-6
node 1001: juju-3f2dbe-21-lxd-7
node 1002: juju-3f2dbe-22-lxd-6
primitive res_ks_6f94244_vip IPaddr2 \
        params ip=172.16.128.9 \
        op monitor timeout=20s interval=10s depth=0
primitive res_ks_haproxy lsb:haproxy \
        op monitor interval=5s
group grp_ks_vips res_ks_6f94244_vip
clone cl_ks_haproxy res_ks_haproxy
property cib-bootstrap-options: \
        have-watchdog=false \
        dc-version=1.1.18-2b07d5c5a9 \
        cluster-infrastructure=corosync \
        cluster-name=debian \
        no-quorum-policy=stop \
        cluster-recheck-interval=60 \
        stonith-enabled=false \
        last-lrm-refresh=1587141039
rsc_defaults rsc-options: \
        resource-stickiness=100 \
        failure-timeout=0

There is no collocation stanza that would enforce running VIP only on node where haproxy is running. Example:
clone cl_ks_haproxy res_ks_haproxy
colocation haproxy_6f94244_vip inf: res_ks_6f94244_vip res_ks_haproxy

[Nikolay Vinogradov (nikolay.vinogradov)]

Testing confirms it would fail and VIP won't migrate if haproxy is killed and not respawned: https://pastebin.canonical.com/p/QR4fRP4CcK/

However automatic haproxy restarting mitigates the issue.