Found by provisioning disposable computes with the bootstrap:
- Ubuntu and Debian: a socket-activated libvirtd ignores libvirtd_opts
"-l", so TCP needs libvirtd-tcp.socket; Ubuntu also kept SASL on
gssapi instead of digest-md5.
- EL10: only virtqemud and virtproxyd were enabled, so storage, network,
interface and other drivers did not answer; and the SELinux policy lets
virtproxyd read but not map the SASL database, so TCP logins failed
(a one-rule module allows it).
- Run from a clone, the bootstrap installs that clone's gstfsd and unit
instead of downloading them.
The RHEL manual install failed at libvirt-devel, which is in CRB; the
installer already enables it. openSUSE's nginx.conf has its own port-80
server for localhost, which the RHEL steps already warn about. The
bootstrap skipped /etc/os-release as a symlink, which on openSUSE Leap
is the only release file, so it found no distribution.
The compute bootstrap installs gstfsd and webvirt-gstfsd.service in one
helper and no longer installs or restarts supervisor; the README's
compute step does the same, a remote panel's bind goes into a drop-in,
and a compute moving from supervisor removes its gstfsd program.
echodebug compared _ECHO_DEBUG and BS_TRUE with -eq, but neither is set
anywhere, so each step printed "[: : integer expression expected". It
now prints debug lines only when _ECHO_DEBUG=1 is set.
The quoting of the tr argument in the /etc/os-release parsing put the
closing parenthesis inside a single-quoted string, so sh and bash
stopped with "unexpected EOF while looking for matching )" before doing
anything: 'curl ... | sudo sh' failed on every distribution. It is now
tr -d "\"'"; bash -n and shellcheck -S error pass.
Also drops the spice_listen edits of qemu.conf; only VNC is used.
