-
Type:
Bug
-
Resolution: Incomplete
-
Priority:
Major - P3
-
None
-
Affects Version/s: 1.6.4
-
Component/s: Sharding
-
None
-
Environment:CentOS release 5.4 (Final) mongodb-linux-x86_64-1.6.4.tgz
-
Linux
-
None
-
3
-
None
-
None
-
None
-
None
-
None
-
None
The machine on which the corrupted mongos runs was under heavy load (some operations like 'compressing files' ). In the mean time, some data was written into the cluster ( some of them are common collections, along with GridFS ). By the way, the written was through another mongos (running on another machine). Also worthy to note that, there was one config-server running on the same machine as the corrupted mongos.
After some time, connection through this mongos couldn't find some of the GridFS files which was written recently while the other mongos could. But it could still find some recent written data of common collections.
Killed the corrupted mongos, started it again, then the difference went away.