Cirrus-CI: retry pkg installation on failure

[emaste]

Summary

Pkg installation failed reasonably frequently, always at:

[62/104] Fetching jpeg-turbo-2.0.6.txz: .......... done
pkg: http://pkgmir.geo.freebsd.org/FreeBSD:13:amd64/quarterly/All/jbigkit-2.1_1.txz: No route to host

Move pkg installation to a script and retry once upon failure.

[emaste]

.cirrus-ci/pkg-install.sh

14

stray "

[emaste]

.cirrus-ci/pkg-install.sh

4

perhaps this list should continue to live in .cirrus.yml and be passed to this script -- will update

[emaste]

Put pkg list back in .cirrus.yml

[emaste]

A version of this worked:
https://cirrus-ci.com/task/6615935889965056

pkg: http://pkgmir.geo.freebsd.org/FreeBSD:13:amd64/quarterly/All/jbigkit-2.1_1.txz: No route to host
pkg install failed

dmesg tail:
random: unblocking device.
intsmb0: <Intel PIIX4 SMBUS Interface> irq 9 at device 1.3 on pci0
intsmb0: Could not allocate I/O space
device_attach: intsmb0 attach returned 6
vtrnd0: <VirtIO Entropy Adapter> on virtio_pci2
random: registering fast source VirtIO Entropy Adapter
lo0: link state changed to UP
debugnet_any_ifnet_update: Bad dn_init result from vtnet0 (ifp 0xfffff80003eff800), ignoring.
debugnet_any_ifnet_update: Bad dn_init result from vtnet0 (ifp 0xfffff80003eff800), ignoring.
vtnet0: link state changed to UP

trying again
Updating FreeBSD repository catalogue...

dmesg snippet looks like messages from boot though, so probably no link flap at this time
debugnet_any_ifnet_update: Bad dn_init result from vtnet0 (ifp 0xfffff80003eff800), ignoring. are curious

[jrtc27]

.cirrus-ci/pkg-install.sh

3

Maybe set -e for future safety? As it stands the script is fine wrt exit codes, but seems wise to add it.

[imp]

@jrtc27 is right for the long term...

[jrtc27]

In D30613#687823, @imp wrote:

@jrtc27 is right for the long term...

As much as I want to say "this would ideally be relatively temporary" we all know how temporary things end up... :)