***1) Preparatory steps***

* Add container tagging with a hash commit so we can always understand which version an error occurred on and roll back to the previous one
* Configure cleaning of old images from the registry

***2) Create a separate gear network for testing using gear-loader***

* Make new n vms in amazon

*Question for Sabaun and Dmitry:*

What kind of network do we need ?
I mean how many validators and do we need an rpc node?

* Make a separate workflow for deployment to these vms
* Сonnect current monitoring and alerting to this network and create new telegram channel for it
* In alert we need also provide hash commit gear binary
* Install an exporter for logs or configure it through the current alerting when we see error or some panic in the logs

*Question for Sabaun and Dmitry:*
Describe please which errors react to.

* When we get an alert, we need to take a snapshot of the current database and provide the ability to download this snapshot

***3) Sending an alert from gear-loader***

*Suggestion for Sabaun and Dmitry:*

I propose to add to the gear-loader the ability to send alerts directly to telegram.
Enough code examples on the internet.
Should be possible when starting gear-loader pass through env $TelegramToken and $chatID