For Tyk for Enterprise On-prem, I believe the configuration metadata for the Tyk cluster is stored in MongoDB. In circumstances where the whole MongoDB cluster is down for a period of time (eg. DBA deleted a critical MongoDB collection by accident), can Tyk servers in the Cluster still start-up and function correctly even if they can’t reach MongoDB for the configuration metadata?
Thanks
P
Hi @pkdon,
The Dashboard currently depends on MongoDB and though there are long-term plans to eventually remove this dependency, for the time being, it won’t be possible to compile the Dashboard whilst your MongoDB server is down.
Kind regards,
Jess @ Tyk
Tyk gateways, when configured to be part of a Multi Data Center Bridge (MDCB) configuration (this is an enterprise component), will continue to serve traffic to known clients even if a scaling event occurs while back-end (Dashboard, mongo, master redis keystore) are down. A scaling event would essentially cause a “dumb” gateway to start, and if it cannot reach the back-end it will pull the last good configuration from the local redis cache.
MDCB encourages the use of many slaved gateway clusters with localised redis caches served my a master mongo/redis/Dashboard system that is the ultimate configuration source.
This is the most resilient configuration available.
A standard Pro installation will not do this - it will require a dashboard and mongo instance to be available for a new gateway to start,