vSphere Replication Server Takes Longer to Connect to vCenter Server

Products

• VMware Live Recovery
• VMware vSphere ESXi

Issue/Introduction

Symptoms:

• Reconnecting to the vSphere Replication server to vCenter Server manually takes longer time.
• You see this error in the vSphere Web Client:
  
  The "Reconnect the selected vSphere Replication Server." operation failed for the entity with the following error message.
Cannot determine the network address of the Virtual Machine.

Environment

vSphere Replication 8.x

Cause

This issue arises because the denylist/allowlist for vSphere Replication is not implemented, preventing effective management of which hosts can be connected to for adding datastores in a vCenter datacenter. As a result, the system may attempt to connect to hosts that cannot be reached, leading to increased connection times.

When the hbrsrv process encounters hosts that it cannot connect to, it may take approximately 20 minutes or more to cycle through all hosts before beginning to load groups from the database. For each group, the following tasks are performed:

1. Create replica image
2. Refresh instance
3. Load instance from database
4. Load image
5. Configure datastore
6. Add datastore to storage map

For a functioning host, adding a datastore is relatively quick, as indicated by messages like:

2017-05-04T13:38:00.933-04:00 [7F028A6AD720 verbose 'StorageMap' opID=hs-init-2090844] Host host-1058 added for datastore 52fxxxed-a4xxxx37-xxxx-3440xxxxxf18
2017-05-04T13:38:00.933-04:00 [7F028A6AD720 verbose 'StorageMap' opID=hs-init-2090844] Host host-13055 added for datastore 52fxxxed-a4xxxx37-xxxx-344xxxxx4f18

However, for problematic hosts, hbrsrv schedules connection retries, which must wait for connection timeouts. If the add datastore task encounters a bad host during this retry, it cannot obtain the host's lock, causing additional delays. The cumulative effect can lead to significantly longer startup times, especially as the number of bad hosts or groups increases.

In theory, the startup time for the hbrsrv service would be reduced if all hosts could be connected to without issues.

Resolution

To work around this issue, you can adjust the following settings:

1. Open the /etc/vmware/hbrsrv.xml file using a text editor.
2. Decrease the hostConnRetriesForLastAddress value.
3. Increase the hostCheckConnectionIntervalSec and hostReconnectIntervalSec values to reduce the frequency of connection attempts.

These adjustments can help optimize the connection process and improve overall performance.