Anyone else seeing a drop with the current 7.x release?
Here is my experience since my last post, but this is just from memory so there’s a bit of missing information:
- Moved from Mongo 4.2 to Mongo 7, custom deployment on an AWS instance with 32GB RAM.
- Large regex col-scans slowed down a significant amount, by about 10x.
- Downgraded to Mongo 6 on the same instance. Col-scans returned to normal speed.
- Migrated from the Mongo 6 instance to Mongo 8 on Mongo Atlas (M30 tier).
- Large regex col-scans slowed down about 1.5x, which was acceptable for my use-case.
Because I migrated to Atlas, I can’t say if my current performance is due to Atlas or Mongo 8. If I were using a custom deployment, I would probably spin up a new test instance using Mongo 8, import all the data, and test the performance. Then just skip straight to Mongo 8 if it performs reasonably.
Mongo 6 community edition was not supported on the latest AWS Linux 2023 so I had to upgrade for SLA reasons.
We gave it another shot and upgraded to 7.0.16, however the problem persisted.
In our case, the issue was that the new dynamic ticket algorithm provided far too few tickets:
(cf. mongo/src/mongo/db/catalog/README.md at master · mongodb/mongo · GitHub )
After the upgrade to 7.0.16 the available read/write tickets drop as low as 8 tickets in comparison to 128 tickets in 6.0.13
We managed to resolve the problem by setting storageEngineConcurrentWriteTransactions
and storageEngineConcurrentReadTransactions
to a fixed value of 128, just as in MongoDB 6. Afterwards the performance also returned to MongoDB 6 levels!
If you’re facing a similar issue, this might be worth trying!
I just upgrade to MongoDB v7.x this morning and users are reporting slow queries.
Are there any findings and solutions for this issues?
Thanks!
Sally
Yep, I experienced the same thing after upgrading from v6 to v7.0.14
Sally
Did you try to adjust storageEngineConcurrentWriteTransactions
and storageEngineConcurrentReadTransactions
? In our case this solved the slow queries issue.