[SERVER-32786] Slow Inserts on MongoDB Sharded Cluster Created: 19/Jan/18  Updated: 06/Dec/22  Resolved: 16/Feb/18

Status: Closed
Project: Core Server
Component/s: Index Maintenance, Performance, WiredTiger
Affects Version/s: 3.6.0
Fix Version/s: None

Type: Question Priority: Minor - P4
Reporter: Abhishek Kulkarni Assignee: Backlog - Triage Team
Resolution: Incomplete Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Assigned Teams:
Server Triage
Participants:

 Description   

I am having issues with progressively worse performance for inserts and Updates on my MongoDB cluster. Initially, I was getting speeds upto 10 cr records in 19 minutes. Now, it is down to 31.30L / hr.

The data configuration is as follows -
1. 8 collections having a cumulative of 30cr records.
2. 1 collection having upwards of 190cr records.

The delay only gets significant on the part 2 (I am getting the speeds of 30 L/hr, on inserts after 160 cr records are already on that collection).

There are 4 indices on that collection. One default on (_id), two on 1 string field each and finally a compound index on 4 fields. The final compound index is being used for sharding.

I am running a MongoDB Sharded cluster, with two shards, each having one primary and two secondary data nodes. I also have a set of three configuration servers (1 primary and 2 secondaries). The nodes are divided as follows

1. Machine 1 -> 1 node from replica set A, one node from replica set B.
2. Machine 2 and Machine 3 -> Same as above.
3. Machine 4,5,6 -> Each having a Configuration Node
4. Machine 7 -> Runs a Mongos (Query Router).

Configuration of the Systems (128GB RAM and 12 cores). Do I need to change anything in terms of the distribution of my nodes across servers and/or any changes in the collection's indexing and sharding?

Any other advice to improve the data insertion and updation is also appreciated.



 Comments   
Comment by Kelsey Schubert [ 16/Feb/18 ]

Hi Kulkarni,

We haven’t heard back from you for some time, so I’m going to mark this ticket as resolved. If this is still an issue for you, please provide additional information and we will reopen the ticket.

Regards,
Kelsey

Comment by Michael Cahill (Inactive) [ 21/Jan/18 ]

Hi Kulkarni, I'm sorry that you are having this issue. To help us diagnose what is causing the inserts to slow down, can you please create an archive of the files in the diagnostic.data directory on each of the servers and attach that information to this ticket?

Generated at Thu Feb 08 04:31:18 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.