What is the problem that needs to be solved? As a user (MongoDB Driver) you'd like to check revocation status of X.509 certificate via TLS Certificate Status Request extension (also known as OCSP stapling) instead of doing outgoing HTTP connection to OCSP endpoint from a user (MongoDB Driver) side.
Why is it a problem? (the pain) MongoDB Server v. 4.4.X introduces OCSP stapling functionality, however OCSP stapling is not enabled on Atlas Clusters with MongoDB Server v. 4.4.X.5 votes
On the Private Endpoint (and VPC Peering) setup screen in Atlas, present options to:
- include or modify "security group"
- attach a "tag = name"
That way when user has the AWS CLI command generated, it includes security & tag1 vote
On Atlas when creating private endpoints, add comments field to allow easier identification.1 vote
enable the user creation to have the following option
clientSource: ["<IP>" | "<CIDR range>", ...],
serverAddress: ["<IP>" | "<CIDR range>", ...]
Live Migrate allows for migrations from non-Atlas clusters into Atlas by syncing data from the source to the Atlas destination, and then cutting over the application to direct workload to the destination.
It's possible to test the application on the destination before cutover, but after cutover, there's no way to sync data in reverse in case unforeseen issues arise and it's necessary to fall back to the source (or another) cluster again.
Having a "fallback sync" option could help with this.7 votes
We have partially older mongoDB versions running on our clusters. We get notified about it by EOL mails from mongoDB. However, as an org owner, I cannot see which clusters are affected. This makes it extremely time consuming for me to identify the relevant contacts per cluster. Our org structure entails self-sufficient teams with independent accountability of the mongoDB clusters. There is no central contact person.
It would be desirable to be able to call an overview of all clusters with the respective installed version in the GUI with owner rights.1 vote
As described in support case #00739901, we have exceeded the 50 limit of clusters in our production project.
We are going to have a lot of new clusters in the coming year, and they all need to be in the production environment.
There should be a way either to add more clusters to one project or to share the same settings between multiple production projects for growing customers of MongoDB like us.2 votes
In Q4 2020, the notimeout cursor was no longer allowed on free or shared instances.
As several DB management tools hardcoded this option for optimum performance it breaks compatibility with them. We use a combination of free/shared-tier instances for dev in combination with dedicated instances for production; this limit effectively means we either have to WAY overprovision our dev resources, or we can't use our tooling anymore on it.
Please provide a way to simply ignore the notimeout option, instead of breaking its connection.
When creating large new clusters sometimes Atlas does not have enough resources available from the cloud provider. When this happens a restore or cluster change fails without reporting what the actual issue is.
Solution: Verify available capacity and issue a warning or error if it is possible that the operation will fail & recommend opening a support ticket (or have a dialog to automatically do so listing the details)2 votes
Some situations require multiple private endpoints (using the Private Link service) to be created per region. This is not currently supported for multi-region replica sets.
One scenario is especially for when there is a shared VPC/VNet for initial transit from on-prem to cloud plus another VPC/VNet for an application. Both of those VPC/VNets would want a Private Endpoint to connect directly to the Atlas cluster. Even if it is possible (sometimes is not), routing traffic via multiple VPC/VNets would mean 3 hops to Atlas from on-prem2 votes
Enable SNMP setup on Atlas Resources.1 vote
Please see details in https://support.mongodb.com/case/006824802 votes
Using the Invoices API to aggregate cluster costs by label involves cross-referencing the Cluster API. Including cluster label information as part of the Invoices response would eliminate this extra step.3 votes
Please allow us to kill user threads on Mongo secondaries. Sometimes we have long running queries on secondaries and need to kill those threads but can't do anything about it since we don't have admin privileges to kill threads.
Also please add the ability to manually restart secondaries, not just failover the primary.
Allow the "SET GLOBAL mongodbmaxvarcharlength = n" to be permanent after a Mongosqld restart. Currently, the setting is ephmeral and must be set everytime the mongosqld daemon is restarted. Currently, our production environment mongosqld can restart without our knowledge at anytime. This results in the mongodbmaxvarharlength variable being reset to zero, which can lead to a production outage. Is there anyway this can be automated, maybe through database trigger?26 votes
A field indicating whether MFA is on for organization users (i.e. on https://docs.atlas.mongodb.com/reference/api/organization-users-get-all-users) would be extremely useful!2 votes
When we upgrade MongoDB there is the possibility that orphaned collections that need to be deleted can prevent the upgrade from completing.
It would make sense for Atlas to include a proactive utility that searched a customers sharded cluster for orphaned docs and gave them the option to clean up the orphaned docs ahead of a major version upgrade.1 vote
To be able to change and read Organizational Settings in an automated matter, support for Organizational Settings should be added to the Mongo Atlas API.3 votes
Atlas should be able to throttle IO in a way that does not allow excessive IO to degrade a node or take a node offline. Current functionality has no defense against a high IO data job other than to allow a node to fail.1 vote
- Don't see your idea?