Installed another ns8 server and joined as cluster node2 to node1
node2 successfuly joined. Its working (no errors displayed)
add LDAP provider to node2 with success
clone apps to node 2 with success
what next to get it working … if node1 is offline ?
Still searching replication settings and cannot find
after turn off node1 … services not working (roundcube + mail)
any tip Im glad to … consume with pleasure
THX for advance
and sorry for trouble
have a nice sunday
NS8 core itself doesn’t provide app replication. Clone and Move are one-shot actions that copy an app’s state at that moment, with no ongoing sync afterward, so cloning mail/Roundcube to node2 gives you a snapshot, not a live replica that takes over if node1 goes down.
Replication, where it exists, comes from the apps themselves — for example the LDAP/Samba AD provider apps support multi-master replication — but mail (Postfix/Dovecot) and Roundcube don’t have that built in, so what you’re seeing is expected behavior, not a misconfiguration.
thx for explaining it so clearly
but … stil 2 question … on my mind
Is it possible to make script doing that: clone and move app … and put it to schedule
Wonder about purpose to make a cluster at all
if You can backup and restore app at any non clustered ns8 machine.
is clustering has other purposes in ns8 ?
Good question. Clustering in NS8 is mainly about managing and running more apps conveniently across more hardware — centralized management from one UI/API, and spreading apps across nodes based on resources — not (yet) about automatic failover.
If you need HA today, that’s better achieved at the VM hypervisor level (e.g. live migration, hypervisor-level replication) rather than trying to replicate it at the NS8 app layer.
Worth noting: back in NS7 there was a feature called “hotsync” that attempted exactly this kind of node-to-node sync/failover, but with virtualization now handling that job well, the idea has become a lot less attractive to pursue again in NS8.
Thx for advice Davide I’m aware about replication in Virtualization apps. For me that one who invent virtualization and bring it to … today level … should get a Nobel price.
Respect for all the people who did it and still doing that.
Sometimes I would like to code a hotsync path for a module. Clone the module once, then keep the data in sync with rsync afterwards. The building block is already there: clone-module runs a transfer-state step that rsyncs the module volumes and the agent state directory to the destination, so the periodic part is mostly a matter of repeating that step instead of the whole clone.
The data side is the easy half though. When the primary goes down and we want to switch to the replica, there is still at least one manual step: repointing the A/AAAA record to the replica node.
That step is precisely what Kubernetes removes. There the DNS name points to a Service VIP that never moves. The EndpointSlice controller drops the failing pod as soon as its readiness probe fails, kube-proxy reprograms the routing on every node, and traffic follows the healthy replica without anyone editing a DNS record. We have no equivalent of that in NS8, so the switch stays manual.
Im not a expert but I think a simple script on router like mikrotik can do the trick with DNS switch after node1 is offline
You can put second trefik path additionaly too
but that script with rsync module and rsync volume … You definitely got mine attention
is it workin with terminal on ns8
Just an idea but yes it is intended to be used in a terminal, probably a python script, anyway it is a global work and the dns issue is the missing piece
