The default runtime
This ladder continues from the node the app ladder’s pages 00 to
07 left running: home01 in learn/home, the Pro licence, the
active tenant prod, and the deployments web, db, pgweb, placed,
store and cloud. Keep them. A runtime belongs to a node, so each page
after this one joins a node that runs its deployments through a different
engine.
A runtime is the engine a node runs deployments through. SHC has four:
docker, podman, incus and apple. Every node has one, chosen at
shc init or shc node join. home01 said nothing and got docker,
because Docker was the engine SHC found on it. Read it.
$ shc node ls --verboseNAME ID HOST INTERNAL UNDERLAY RUNTIME CORES MEMORY FREE CORES FREE MEMORY OUTSIDE RESERVED STATE CONDITION AVAIL CLOUD LOCATION ROOM BUILD LEADER VOTERhome01 node-090212582e76 home01 10.192.0.1:4003 167.233.221.15 docker 2 7.6G 2.0 7.6G - - healthy ready active learn learn/home ok 0.1.2 yes yes$ shc node status...$ shc node info home01...The RUNTIME column reads docker. shc node status is the row of the
node you are on; shc node info is one node’s row with the deployments it
runs and its share of the fabric.
A deployment can name the runtime it needs. One that names none runs on
any. web named none, so SHC put it on a node of its choice, and the one
node there was home01.
$ shc init --runtime podman --nic eth1$ shc cluster init --runtime incus --nic eth1$ shc install ~/learn/nginx -s web --runtime docker$ shc clone web web2 --runtime podman$ shc backup restore <backup> --runtime podman$ shc recover --runtime podmanshc init and shc cluster init take the first node’s runtime the way
shc node join takes a joining node’s, and --nic on any of the three
names the interface the cluster’s own traffic binds on a machine with
more than one. docker is the word for the default engine; the engine
goes by compose inside SHC, and that word is refused on the command
line. A clone, a restore and a recovery each take the runtime the copy
should run on; unsaid, it is the source’s.
$ shc ls...Before page 01, prepare two more Debian 13 machines with 2 CPUs and 8 GB
of memory each, reachable from home01 and from each other on the cluster
ports listed in Installation.
Name them pod01 and inc01. Do not install SHC on them yet: the next
two pages install each one’s engine first and SHC after it. The package
brings Docker to a node that has no engine, and a Docker that has started
once leaves the kernel dropping forwarded traffic until the next boot,
which shc node join refuses with X500396FWHDRP. Give them the learn
nodes’ names, as the learn nodes have, and let root on each ssh home01
with a key and no prompt: a join reads its token from home01 that way. The
licence from the app ladder covers ten nodes, so it covers both.
Next: A podman node.