When terminating a function and replacing it during an update,
there was often an error about task precondition not met
which meant having to try and or wait or being left in an
inconsistent state.
The new flow makes sure "Wait" is called in either code path
and allows for a custom gap between the SIGTERM and SIGKILL
through the grace_period env var - set as a Go duration.
Signed-off-by: Alex Ellis (OpenFaaS Ltd) <alexellis2@gmail.com>
All services like docker and k8s.io use their own namespaces
for core services, this change moves openfaas services into
the openfaas namespace instead of the default one.
The main change is that logs will look like:
journalctl -t openfaas:gateway
Instead of "default:gateway"
Function logs will remain unaffected and scheduled in the
openfaas-fn namespace.
Signed-off-by: Alex Ellis (OpenFaaS Ltd) <alexellis2@gmail.com>
The behaviour prior to this patch caused some confusion for
users since they expected a behaviour like Swarm / Kubernetes
which always pulls images by default, even if cached. I've tested
the change and it is working as expected. By default images are
always pulled upon deployment.
To revert to the prior behaviour, simply add to faasd up:
--pull-policy=IfNotPresent
Signed-off-by: Alex Ellis (OpenFaaS Ltd) <alexellis2@gmail.com>
Fixes a bug when attempting to access a non-existant IP from
GetIPfromPID called via the list API.
Renames the provider from faas-containerd
Updates function deletion grace period to 30s to prevent any
errors in the REST API during a long-running deletion.
Tested on Linux with the figlet function which by default takes
around 5s to delete due to its write_timeout value, the deletion
now blocks rather than throwing an error.
Signed-off-by: Alex Ellis (OpenFaaS Ltd) <alexellis2@gmail.com>
