Mongo cluster over different versions

28 views

Skip to first unread message

pas...@gmail.com

unread,

Aug 16, 2017, 4:14:16 PM8/16/17

to mongodb-user

Hello guys,
can anybody help me with Mongo cluster settings with followed requirements ?

we would create cluster creating between Mongo 3.2.10 on cust.side and 3.4.4 (optional) enterprise local side .
We can't break customer's data availability on customer side, with short service restart only ( max one times per day )

Main idea:
Create sync. replicated cluster between two sides in first step.
When data are replicated online = sync , then web data traffic will be forwarded to our side and we recreate two node cluster on our side then ( original server will be stopped ) .

So, both Mongo versions 3.2.10 versa 3.4.4 have different syntax and really is not possible "kill" customer server for a minute .
Customer's Mongo is simple single instance, but with bilions rows.
Exists any nondesctructive and non-long-interruptable solution for it ??
Regs.
Pavel

Kevin Adistambha

unread,

Aug 24, 2017, 8:37:56 PM8/24/17

to mongodb-user

Hi Pavel

When data are replicated online = sync , then web data traffic will be forwarded to our side and we recreate two node cluster on our side then ( original server will be stopped )

So, both Mongo versions 3.2.10 versa 3.4.4 have different syntax and really is not possible “kill” customer server for a minute .

I’m not certain I understand your goal. What are you trying to do, exactly? Are you trying to move data from a remote deployment to a local deployment? Why do you need different versions?

To answer your direct question, it is possible to have mixed MongoDB versions in a replica set. However, this is supported only for the purpose of upgrading the whole set, and were not tested for a long-running production setup. For the exact procedure, see Upgrade a replica set to 3.4.

It is also possible to create a replica set with nodes in different physical location. However, careful and extensive planning is required before undertaking this major change to a deployment (I understand the “customer” deployment is a standalone node), and I would not recommend you to attempt this before doing thorough testing and taking verified backups. For more details on this matter, see: