-
Bug
-
Resolution: Won't Do
-
Critical
-
None
-
openshift-4.10.z
-
None
The customer has deployed the OCP clusters recently and is facing a "Connection reset by peer (&&) Connection was refused - did you specify the right host or port" error?
- All nodes are deployed in the same network and each location/VMware cluster has 2 OCP4.10 clusters.
- Each VMware cluster has a different network/ switch since all are in other Geo locations.
- Customer has provided the location details for each cluster, where you see 2 OCP clusters only those are in the same VMware cluster and the same network, other than that 1 OCP cluster in an individual cluster and network.
- All cluster has separate LBs servers (API+ Ingress) and 6 nodes (master + worker), so each cluster has 8 VMs running.
- Customer facing the same connection reset issue in all locations, not only in the 2 OCP clusters running location.
As below:
Dallas - 2 ocp
Wdc - 2 ocp
Frankfurt - 2 ocp
Singapore - 1 ocp
Tokyo - 1 ocp
Amsterdam - 1 ocp
Southbury - 1 ocp
Provider: VMware
Installation type: UPI
Network environment connected: connected
Internet Proxy usage: No
-As the customer shared that they noticed one thing:
1. They have configured API LB IP with 9.172.54.29, that same IP address can able to see in one of the master nodes in vCenter.
2. Ingress LV IP with 9.172.54.30, that same IP address can able to see in one of the worker nodes in vCenter.
Because of these IP conflicts, I can't log in to the LBs server through SSH from the terminal, most of the time getting a connection reset error.
- while performing oc commands task from base-station connection reset and connection refused errors are shown below:
~~~~~~~
Example of oc command:
[root@dal-wg-ocp-utility ~]# oc create identity w3ID:udhayt13@in.ibm.com
The connection to the server api.frasos-ocp4.os.net.ibm.com:6443 was refused - did you specify the right host or port?
[root@dal-wg-ocp-utility ~]# oc create useridentitymapping w3ID:udhayt13@in.ibm.com udhayt13@in.ibm.com
The connection to the server api.frasos-ocp4.os.net.ibm.com:6443 was refused - did you specify the right host or port?
[root@dal-wg-ocp-utility ~]#
[root@dal-wg-ocp-utility ~]# oc create identity w3ID:udhayt13@in.ibm.com
The connection to the server api.frasos-ocp4.os.net.ibm.com:6443 was refused - did you specify the right host or port?
[root@dal-wg-ocp-utility ~]#
[root@dal-wg-ocp-utility ~]# oc create identity w3ID:udhayt13@in.ibm.com
The connection to the server api.frasos-ocp4.os.net.ibm.com:6443 was refused - did you specify the right host or port?
[root@dal-wg-ocp-utility ~]#
[root@dal-wg-ocp-utility ~]# oc create identity w3ID:udhayt13@in.ibm.com
Unable to connect to the server: net/http: TLS handshake timeout
[root@dal-wg-ocp-utility ~]# oc patch kubecontrollermanager cluster p='{"spec": {"forceRedeploymentReason": "recovery'"$( date )"'"}}' --type=merge
The connection to the server api.wdcsos-ocp4.os.net.ibm.com:6443 was refused - did you specify the right host or port?
============
[root@api ~]#
[root@api ~]# client_loop: send disconnect: Broken pipe
udhayakumart@udhayakumars-MacBook-Pro ~ %
==============
~~~~~~~~
- OCP connection is stable due to this API issue and the console is not loading sometimes and the must-gather is also not getting completed.
- account is impacted by
-
OCPBUGS-7411 On newly deployed VMware UPI cluster the IP address conflicting with LBs and Master/Worker nodes
- Closed