Hi, Please properly subscribe to the mailing list so that the community can receive email notifications for your messages. To subscribe, send empty email to [email protected] and follow simple instructions in the reply.
vahan wrote > Imagine we have cluster of nodes with replicated cache. Each node is > writing something to that cache. > Then suddenly the connection of some node is lost, that node is still > running and continuing to write a data to his own copy of the cache. When > connection will back to online again what would happen? Will the locally > written data be synched up with other nodes' cashe (will the information > written during connection lost available in the replicated cashe) ? Ignite does not allow to merge clusters after split-brain, because this will most likely cause data inconsistency. You will end up with two separate clusters and will have to manually stop and restart one of them. It will then join other cluster (assuming that network is restored) and rebalance the data. Updates happened on the stopped cluster within this period of time will be lost. BTW, GridGain provides better support for segmentation handling as a part of their paid solution [1]. Basically, they allow to detect segmentation automatically and stop/restart one of the clusters right away, so there are no time frame with two clusters and therefore no lost updates. [1] https://gridgain.readme.io/docs/network-segmentation -Val -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/How-can-I-save-the-job-result-which-was-ran-on-server-which-lost-the-connection-tp9462p9612.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
