Troubleshooting Cluster Issues

Cluster Creation

Public Cloud Provider

• Make sure the permissions for the account you provided to PMK as part of cloud provider creation has all the required privileges. See the AWS pre-requisites under Getting Started section for more details

Cluster Creation Fails for BareOS

• Navigate to Infrastructure -> Clusters tab.
• Click on the cluster name. This will take you to the cluster details page.
• Click on the “Node Health” tab

Here you should see detailed breakdown of which nodes failed to install and which specific steps failed. Next, check Troubleshooting Node Issues.

Etcd

Heartbeat/Election Timeout Interval

2021-02-04 18:36:31.380207 W | etcdserver: failed to send out heartbeat on time (exceeded the 100ms timeout for 124.999498ms, to 92d6e239c543436)

2021-02-04 18:36:31.380220 W | etcdserver: server is likely overloaded

2021-02-04 18:36:31.382208 W | etcdserver: read-only range request "key:\"/registry/mutatingwebhookconfigurations/vault-agent-injector-cfg\" " with result "range_response_count:1 size:2723" took too long (264.355727ms) to execute

ETCD_HEARTBEAT_INTERVAL - This is the frequency with which the leader will notify followers that it is still the leader.

ETCD_ELECTION_TIMEOUT - This timeout is how long a follower node will go without hearing a heartbeat before attempting to become a leader itself.

By default, etcd uses a100msheartbeat interval and1000mselection timeout.

# cat /etc/pf9/kube.env | grep -i etcd

export ETCD_HEARTBEAT_INTERVAL="1000"

export ETCD_ELECTION_TIMEOUT="10000"

Database Size Exceeded

etcdserver: failed to apply request,took 2.429<C2><B5>s,request header:<ID:1920634987875929770 > txn:<compare:<target:MOD key:"/registry/services/endpoints/kube-system/kube-controller-manager" mod_revision:287319046 > success:<request_put:<key:"/registry/services/endpoints/kube-system/kube-controller-manager" value_size:473 >> failure:<>>,resp ,err is etcdserver: no space

1. Stop the pf9-hostagent and nodeletd services on the master node(s).

sudo systemctl stop pf9-{hostagent,nodeletd}

2. Issue a stop for the Nodelet phases.

/opt/pf9/nodelet/nodeletd phases stop

3. In /opt/pf9/pf9-kube/master_utils.sh , modify the function ensure_etcd__r_unning()to add the following environment variable.

--volume ${ETCD_DATA_DIR}:/var/etcd/data \

        -e ETCD_DEBUG=${DEBUG}

        -e ETCD_QUOTA_BACKEND_BYTES=<size_in_bytes>"

4. Start the pf9-hostagent service.

sudo systemctl start pf9-hostagent

5. Verify the size was correctly set by scraping the etcd metrics endpoint.

curl -L http://localhost:2379/metrics | grep etcd_server_quota_backend_bytes