-
Type: Bug
-
Resolution: Done
-
Priority: Critical - P2
-
None
-
Affects Version/s: 2.0.6, 2.0.7
-
Component/s: Replication
-
Labels:None
-
Environment:CentOS release 5.5 (Final)
mongodb-linux-x86_64-2.0.6.tgz
mongodb-linux-x86_64-2.0.7.tgz
-
Linux
Hi all
There is a replica set with 2 mongod nodes and 1 arbiter. It is 2000 average updates on master node by mongostat.
All of mongodb oplog: 10g
Master syncdelay: 100s
Secondary syncdelay: 150s
At the request peak, secondary node reports "syncing to: 10.136.24.88:27032" on MongoDB web interface. A monment later, it reports "too stale to catch up". I found that some informations from secondary log, but the network is well because all of nodes are at LAN.
Mon Aug 27 13:10:37 [rsSync] Socket recv() timeout 10.136.24.88:27032
Mon Aug 27 13:10:37 [rsSync] SocketException: remote: 10.136.24.88:27032 error: 9001 socket exception [3] server [10.136.24.88:27032]
Mon Aug 27 13:10:37 [rsSync] DBClientCursor::init call() failed
Mon Aug 27 13:10:38 [rsSync] replSet syncing to: 10.136.24.88:27032
Also I found that the optime of secondary oplog would not be updated when secondary node reports "syncing to: 10.136.24.88:27032" on MongoDB web interface.
I tested MongoDB 2.0.6/2.0.7 to reproduct it.