vault unit moves between ready and error repeatedly

We have a 3 node ha vault cluster, the two slave nodes switch between active (Unit is ready) and
error (hook failed: "update-status") states repeatedly (roughly every 10 minutes)

The associated error message in the logs is:

2018-07-02 16:38:10 DEBUG update-status active
2018-07-02 16:38:10 DEBUG update-status Traceback (most recent call last):
2018-07-02 16:38:10 DEBUG update-status File "/var/lib/juju/agents/unit-vault-2/charm/hooks/update-status", line 19, in <module>
2018-07-02 16:38:10 DEBUG update-status main()
2018-07-02 16:38:10 DEBUG update-status File "/var/lib/juju/agents/unit-vault-2/.venv/lib/python3.5/site-packages/charms/reactive/__init__.py", line 82, in main
2018-07-02 16:38:10 DEBUG update-status hookenv._run_atexit()
2018-07-02 16:38:10 DEBUG update-status File "/var/lib/juju/agents/unit-vault-2/.venv/lib/python3.5/site-packages/charmhelpers/core/hookenv.py", line 1128, in _run_atexit
2018-07-02 16:38:10 DEBUG update-status callback(*args, **kwargs)
2018-07-02 16:38:10 DEBUG update-status File "/var/lib/juju/agents/unit-vault-2/charm/reactive/vault_handlers.py", line 585, in _assess_status
2018-07-02 16:38:10 DEBUG update-status application_version_set(health.get('version'))
2018-07-02 16:38:10 DEBUG update-status File "/var/lib/juju/agents/unit-vault-2/.venv/lib/python3.5/site-packages/charmhelpers/core/hookenv.py", line 970, in application_version_set
2018-07-02 16:38:10 DEBUG update-status subprocess.check_call(cmd)
2018-07-02 16:38:10 DEBUG update-status File "/usr/lib/python3.5/subprocess.py", line 576, in check_call
2018-07-02 16:38:10 DEBUG update-status retcode = call(*popenargs, **kwargs)
2018-07-02 16:38:10 DEBUG update-status File "/usr/lib/python3.5/subprocess.py", line 557, in call
2018-07-02 16:38:10 DEBUG update-status with Popen(*popenargs, **kwargs) as p:
2018-07-02 16:38:10 DEBUG update-status File "/usr/lib/python3.5/subprocess.py", line 947, in __init__
2018-07-02 16:38:10 DEBUG update-status restore_signals, start_new_session)
2018-07-02 16:38:10 DEBUG update-status File "/usr/lib/python3.5/subprocess.py", line 1490, in _execute_child
2018-07-02 16:38:10 DEBUG update-status restore_signals, start_new_session, preexec_fn)
2018-07-02 16:38:10 DEBUG update-status TypeError: Can't convert 'NoneType' object to str implicitly

The error seems to originate because /var/lib/juju/agents/unit-vault-2/charm/lib/charm/vault.py line 209 gets an invalid reponse:
    response = requests.get(VAULT_HEALTH_URL.format(vault_addr=VAULT_LOCALHOST_URL))

I added some print statements into the charm and it is connecting to http://127.0.0.1:8220/v1/sys/health but the response it receives is {'errors': []}

But when I try it by hand (on a machine showing the error) I get:
root@juju-3732e2-18-lxd-14:~# curl http://127.0.0.1:8220/v1/sys/health
{"initialized":true,"sealed":false,"standby":false,"replication_performance_mode":"disabled","replication_dr_mode":"disabled","server_time_utc":1530554818,"version":"0.10.1","cluster_name":"vault-cluster-c5a6c830","cluster_id":"916d5cca-98c5-1e53-23d5-90d13c5bc5c5"}

The leader of the cluster appears to be stable