If we run a simple query on a sharded collection it failes after the reads have finished the first shard.

we can reproduce it with:

Javascript test query as file "script":
function logi(pub) {
print(new Date());
print(pub.someId);
};

db.collection.find().forEach(logi);

log output from this script with command "mongo refind < script ; date":
Wed Mar 21 2012 17:21:23 GMT+0000 (UTC)
NumberLong(1836664)
Wed Mar 21 2012 17:21:23 GMT+0000 (UTC)
NumberLong(1836859)
Wed Mar 21 2012 17:21:23 GMT+0000 (UTC)
NumberLong(1836883)
Wed Mar 21 2012 17:21:23 GMT+0000 (UTC)
undefined
getMore: cursor didn't exist on server, possible restart or timeout?
Wed Mar 21 17:21:23 UTC 2012

log output router which we connect to:
Wed Mar 21 17:21:23 [conn11974] ~ScopedDbConnection: _conn != null
Wed Mar 21 17:21:23 [conn11974] end connection 127.0.0.1:12880