mirror of https://github.com/status-im/consul.git
Merge pull request #15032 from hashicorp/docs/mgw-primary-upgrade
This commit is contained in:
commit
ccf15f4153
|
@ -184,3 +184,18 @@ expected result:
|
||||||
- Ensure any API request that activates datacenter request forwarding. such as
|
- Ensure any API request that activates datacenter request forwarding. such as
|
||||||
[`/v1/catalog/services?dc=<OTHER_DATACENTER_NAME>`](/api-docs/catalog#dc-1)
|
[`/v1/catalog/services?dc=<OTHER_DATACENTER_NAME>`](/api-docs/catalog#dc-1)
|
||||||
succeeds.
|
succeeds.
|
||||||
|
|
||||||
|
### Upgrading the primary gateways
|
||||||
|
|
||||||
|
Once federation is established, secondary datacenters will continuously request
|
||||||
|
updated mesh gateway addresses from the primary datacenter. Consul routes the requests
|
||||||
|
through the primary datacenter's mesh gateways. This is because
|
||||||
|
secondary datacenters cannot directly dial the primary datacenter's Consul servers.
|
||||||
|
If the primary gateways are upgraded, and their previous instances are decommissioned
|
||||||
|
before the updates are propagated, then the primary datacenter will become unreachable.
|
||||||
|
|
||||||
|
To safely upgrade primary gateways, we recommend that you apply one of the following policies:
|
||||||
|
- Avoid decommissioning primary gateway IP addresses. This is because the [primary_gateways](/docs/agent/config/config-files#primary_gateways) addresses configured on the secondary servers act as a fallback mechanism for re-establishing connectivity to the primary.
|
||||||
|
|
||||||
|
- Verify that addresses of the new mesh gateways in the primary were propagated
|
||||||
|
to the secondary datacenters before decommissioning the old mesh gateways in the primary.
|
||||||
|
|
Loading…
Reference in New Issue