Ask Daniel for advice.
Only one of the four mongos can be started. If other mongos are started, all queries cannot be executed.
Tue Aug 13 14:17:16 [conn10647] DBClientCursor::init call() failed
Tue Aug 13 14:17:16 [conn2937] ns: user.dev could not initialize cursor across all shards because : stale config detected for ns: user.dev ParallelCursor::_init @ s02/ 192.168.1.101:9100,192.168.1.102:9100,192.168.1.103:9100 attempt: 0
Only one can be started. Does it mean that only one mongos can be started for normal query? What does it mean that a query cannot be executed? Start and die? Which mongos have this problem? What is the version number, several mongos, several mongos, config server? Was it okay before? What changes have you made?
If one (and at most only one) mongos migrates data between mongd, then other mongos look for the original shard according to the config information of cache, and find that the data is no longer there and config has changed, then staleconconfig will report an error, which is normal. Mongos will go to config server for the new configs, should be able to get the latest configs, find the data, and continue to work happily.