In a replica set, I run mongo import twice a day on primary (spaced out - 12 hours) - which upserts 290k records, causing oplog window to fall below 1 hour. It automatically resolves as write frequency is nominal except these two times a day. Should I increase oplog size to keep the window higher?
How does oplog entries work, do the old ones get removed after they are successfully sent and run on secondary mongod-s? Any info on that would also be helpful
No, check above manual for more information. (Basically rule is: 1. more space needed for new oplog entries and 2. the retention period has passed)
I can’t think of any harm (other than slightly more disk use on oplogs) of doing that.
One related thing is possibly flow control. You can tune it a bit so the oplogs are not filled too quickly.
I only find this help doc for increasing (resizing without restart) oplog size, and I am having trouble connecting to secondaries. Is there a way to do the resizing via GUI? please let me know - also pls let me know, what is the URI for connecting to secondary and primary - where we can get it from the atlas portal?
What tier is your cluster. Some features are not available on the shared tiers: M0,M2,M5