If you use auto-downing and that triggers because of a network partition there will be two separate clusters and they will not recombine automatically. That's why we recommend against using auto-downing and instead recommend something like the Split Brain Resolver http://doc.akka.io/docs/akka/rp-16s01p03/scala/split-brain-resolver.html
/Patrik fre 18 mars 2016 kl. 08:51 skrev 'Christian Hoffmeister' via Akka User List <akka-user@googlegroups.com>: > So basically the known seed nodes are replaces with a known etcd cluster > for the information sharing, if I understand this right? > > Somewhat like this? > > * A joins, no one is there, so it puts inself into etcd and starts a > single node cluster > * B joins, finds A in etcd and joins A > * C joins, finds A.... > * A gets shut down > * After a timeout, B or C are put into etcd as new seed node > * D joins, finds B or C.... > > In the examples ( > https://github.com/rkrzewski/akka-cluster-etcd/blob/master/examples/cluster-monitor/src/main/resources/application.conf) > I found that auto-downing is used. What happens in case of temporory > network separation? For example: > > * A is in etcd as seed node > * A/B (can access each other, but neither etcd nor C/D) > * C/D (can access each other, but not A/B) > > What happens now? And what happen when network is fully up again and > A/B/C/D/etcd can communicate again. Do the two isle recombine into a single > cluster? > > Greetings > /c > > Am Freitag, 18. März 2016 00:13:51 UTC+1 schrieb Christian Hoffmeister: >> >> Hello, >> >> I am just starting to dive into akka-cluster and have a question >> regarding the seed nodes: >> >> My test project consists of 4 projects so far: >> >> * PROTOCOL contains messaging case classes >> * API contains cluster node, that also exposes a rest api (akka-cluster + >> akka-http) >> * AUTH contains cluster node for auth stuff (akka-cluster) >> * USERS contains cluster node to manage user data (akka-cluster) >> >> At the moment I just set one API and one AUTH instance as seed nodes >> (could also have been some other nodes). Is it good practice to have >> another special project (called SEED), that does not do anything in >> particular, >> except for joining the cluster, to act as seed node? >> >> From my first thoughts, this might be a good idea, since this nodes would >> have to be restarted less often then other nodes (that have business >> logic). Basically only when updating infrastructure like host machine or >> akka. >> >> Am I getting something wrong here? >> >> Greetings >> Christian >> > -- > >>>>>>>>>> Read the docs: http://akka.io/docs/ > >>>>>>>>>> Check the FAQ: > http://doc.akka.io/docs/akka/current/additional/faq.html > >>>>>>>>>> Search the archives: https://groups.google.com/group/akka-user > --- > You received this message because you are subscribed to the Google Groups > "Akka User List" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to akka-user+unsubscr...@googlegroups.com. > To post to this group, send email to akka-user@googlegroups.com. > Visit this group at https://groups.google.com/group/akka-user. > For more options, visit https://groups.google.com/d/optout. > -- >>>>>>>>>> Read the docs: http://akka.io/docs/ >>>>>>>>>> Check the FAQ: >>>>>>>>>> http://doc.akka.io/docs/akka/current/additional/faq.html >>>>>>>>>> Search the archives: https://groups.google.com/group/akka-user --- You received this message because you are subscribed to the Google Groups "Akka User List" group. To unsubscribe from this group and stop receiving emails from it, send an email to akka-user+unsubscr...@googlegroups.com. To post to this group, send email to akka-user@googlegroups.com. Visit this group at https://groups.google.com/group/akka-user. For more options, visit https://groups.google.com/d/optout.
