Just wanted to ask if its really affecting the transactions being process on the primary if a secondary node is offline / unreachable status.
Currently when our DR site is down which is part of the replica set of PRODUCTION environment for replication, transactions monitored tend to long run causes slow down on the application. After removing the inactive nodes , transaction runtime became normal.
Is it possible to prevent the slowdown without removing the members. the DR servers have already a priority 0 and voting 0 configuration.
Hoping you can give me some insights to resolve this one. Thanks !
Could you clarify what you mean by “transaction” in this context?
MongoDB supports multi-document transactions starting from the 4.0 series, but you mentioned that you’re using MongoDB 2.6.12, which doesn’t have this feature.
If you are still using MongoDB 2.6, this version is very outdated (released in March 2016) and not supported anymore (e.g. not receiving updates and fixes). Is it possible for you to move to a supported version as soon as possible?
Unfortunately due to the massive difference between MongoDB 2.6 and the latest ones (i.e. MMAPv1 vs. WiredTiger, replica set Protocol Version 0 vs. Protocol Version 1, just to name a couple), it’s difficult to determine what went wrong there (if there is anything wrong with it at all).
Having said that, there are some troubleshooting tips that works pretty much universally, such as:
Are you seeing an increase in log activities (both MongoDB and application logs) in the primary during this period? What activities are most prominent?
Are these slowdowns always been there, or happened just recently? Was something changed in the application or the database deployment setup?
Is it possible that the application is the cause of the slowdown? What happened if you connect directly to the replica set (using a small test application, or using the mongo shell), and try to perform some operations with all nodes online and some nodes offline? Are you seeing the same slowdown patterns?
Hopefully those will help you pinpoint some patterns/causes of the slowdowns.