Hi,
I have an Ignite cluster with 10nodes. Here is the snippet from one of the ignite server log. >>> [13:03:54] Topology snapshot [ver=11, servers=10, clients=0, CPUs=352, >>> heap=1000.0GB] Now I created a cache with mode PARTITIONED and loaded some data into the cache without backups/replication. Here is the snippet from the visor console. visor> cache Time of the snapshot: 11/14/17, 13:08:22 +=====================================================================================================================================================+ | Name(@) | Mode | Nodes | Entries (Heap / Off-heap) | Hits | Misses | Reads | Writes | +=====================================================================================================================================================+ | 936ab7a0-60d3-11e7-b322-8f31ade02ac0(@c0) | PARTITIONED | 10 | min: 112076 (0 / 112076) | min: 0 | min: 0 | min: 0 | min: 0 | | | | | avg: 118185.90 (0.00 / 118185.90) | avg: 0.00 | avg: 0.00 | avg: 0.00 | avg: 0.00 | | | | | max: 124707 (0 / 124707) | max: 0 | max: 0 | max: 0 | max: 0 | +-----------------------------------------------------------------------------------------------------------------------------------------------------+ Now I Intentionally killed one server node. Then the number of servers goes down in server logs which is expected. >>> [13:11:50] Topology snapshot [ver=46, servers=9, clients=0, CPUs=312, >>> heap=900.0GB] Its clear that at this point some data is lost as replication/backups is not enabled. Visor console confirms the same. visor> cache Time of the snapshot: 11/14/17, 13:16:13 +=====================================================================================================================================================+ | Name(@) | Mode | Nodes | Entries (Heap / Off-heap) | Hits | Misses | Reads | Writes | +=====================================================================================================================================================+ | 936ab7a0-60d3-11e7-b322-8f31ade02ac0(@c0) | PARTITIONED | 9 | min: 112076 (0 / 112076) | min: 0 | min: 0 | min: 0 | min: 0 | | | | | avg: 117848.33 (0.00 / 117848.33) | avg: 0.00 | avg: 0.00 | avg: 0.00 | avg: 0.00 | | | | | max: 124707 (0 / 124707) | max: 0 | max: 0 | max: 0 | max: 0 | +-----------------------------------------------------------------------------------------------------------------------------------------------------+ Now I launch a client to read the data from cache and save it to filesystem. The client job doesn't throw any exception/error regarding the server loss. It just reads the data from the servers, saves it to filesystem and exits. Is this behavior expected? Shouldn't the client be aware that there is data loss in the cache ? If this is expected behavior, then how will the client application know about the data loss? In my scenario, I can't have replication enabled but my client application need to read either full data or fail if there is data loss. Is there any configuration for that? Thanks, Pradeep -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/
