As per https://apacheignite.readme.io/docs/partition-loss-policies this mode never marks a lost partition as lost, pretending that no partition
loss has happened and clearing the partition loss state right away.
Technically, the partition will not be added to the collection of
lostPartitions which is the main difference from READ_WRITE_ALL mode.
From the API doc explanation, it looks like that we might get inconsistent
result on different nodes by reading from a previously lost partition.
While from standard docs on website, this is not clearly defined if its
safe, That weather we will consistently just get null on reading from a
previously lost on all nodes or it will be some inconsistent behavior.
Can we please get confirmation which interpretation make more sense ?
Looks like next part is missing for IGNORE policy description in readme.io
"The result of reading from a lost partition is undefined and may be
different on different nodes in the cluster."
There is no guarantee for IGNORE policy that reading from the lost partition
will return "null" on all nodes.