Detail
From the health -v command in asadm, we get a warning saying "Namespace storage under configured for single node failure."This is based on a health rule in our asadm health command.
Answer
/* Following query collects used device space and total device space and computes available free space on each node per namespace per cluster (group by CLUSTER, NAMESPACE, NODE). It collects cluster-size and uses it to find out expected data distribution for each node in case that node is down. It checks max of this computed value per namespace with available space per node per namespace. */ t = select "data_total_bytes" as "disk_space", "pmem_total_bytes" as "disk_space", "device_total_bytes" as "disk_space", "device-total-bytes" as "disk_space", "total-bytes-disk" as "disk_space" from NAMESPACE.STATISTICS; u = select "data_used_bytes" as "disk_space", "pmem_used_bytes" as "disk_space", "device_used_bytes" as "disk_space", "used-bytes-disk" as "disk_space" from NAMESPACE.STATISTICS; /* Available extra space */ e = do t - u; e = group by CLUSTER, NAMESPACE, NODE do SUM(e) save as "available storage space"; s = select "cluster_size" as "size" from SERVICE; n = do MAX(s); n = do n - 1; /* Extra space need if 1 node goes down */ e1 = do u / n; e1 = group by CLUSTER, NAMESPACE do MAX(e1) save as "distribution share of used storage space per node"; r = do e > e1; ASSERT(r, False, "Namespace storage under configured for single node failure.", "OPERATIONS", WARNING, "Listed namespace[s] does not have enough storage space configured to deal with increase in data per node in case of 1 node failure. Please run 'show statistics namespace like bytes' to check storage space. It is non-issue if single replica limit is set to larger values, i.e if number of replica copies are reduced in case of node loss.", "Namespace single node failure storage space config check."); SET CONSTRAINT VERSION < 7.0.0; /* Same as above query but for memory */ t = select "memory-size" as "mem" from NAMESPACE.CONFIG; u = select "used-bytes-memory" as "mem", "memory_used_bytes" as "mem" from NAMESPACE.STATISTICS; /* Available extra space */ e = do t - u; e = group by CLUSTER, NAMESPACE, NODE do SUM(e) save as "available memory space"; s = select "cluster_size" as "size" from SERVICE; n = do MAX(s); n = do n - 1; /* Extra space need if 1 node goes down */ e1 = do u / n; e1 = group by CLUSTER, NAMESPACE do MAX(e1) save as "distribution share of used memory space per node"; r = do e > e1; ASSERT(r, True, "Namespace under configured (memory) for single node failure.", "OPERATIONS", WARNING, "Listed namespace[s] does not have enough memory space configured to deal with increase in data per node in case of 1 node failure. Please run 'show statistics namespace like bytes' to check memory space. It is non-issue if single replica limit is set to larger values, i.e number of replica copies reduce.", "Namespace single node failure memory config check.");