-
Type:
Bug
-
Resolution: Done
-
Priority:
Major - P3
-
Affects Version/s: 2.2.0-rc1
-
Component/s: Performance
-
None
-
Environment:MongoDB 2.0.1 replicaSet with one host "upgraded" (ran mongod from 2.2.0-rc1 linux 64bit download package) to 2.2.0-rc1 for testing. 100GB RAM, RAID 10 spinning disk, CentOS 5.5, Linux 2.6.18-194, Xeon X5675@3.07Ghz, NUMA disabled. Existing 217GB dataset from 2.0.1 with 51GB indexes/
-
Linux
-
None
-
3
-
None
-
None
-
None
-
None
-
None
-
None
-
None
I upgraded one server in a 6-server 2.0.1 replicaSet to test MongoDB 2.2.0-rc1. I wanted to test read and locking performance so I switched to a db with a very large collection that easily exceeds RAM. I then ran a query that I knew would match no records: db.collection.find(
{x: 1}). This caused the mongostat I was running in another window to stop responding, I could connect with another shell but "show dbs" would not return until the above query completed. Also, hitting Ctrl-C on the find() query caused "Fri Aug 17 19:07:19 Assertion: 13111:field not found, expected type 2" in the shell. I was unable to manually kill the find() operation.
I also tried this a second time so I could be sure that as much of the database that could be hot in memory was available.