one_env missing data, context.sh missing

[tkald]

On 1st run, context files are not correctly created:
one_env files has only 3 lines on it (instead of 50) and context.sh file is missing.

cat /tmp/one_env
export TOKENTXT=""
export VMID="2580"
export ONEGATE_ENDPOINT="http://1.2.3.4:5030"

After vm reboot (second run), all seems fine though.

One-context 5.4.1 on Ubuntu 16.04 vm, OpenNebula 5.4.1.
All is working as expected with one-context version 5.0.3.

See attached log:
context5.4.1.log.txt

[vholer]

Hi @tkald, in 5.4 the contextualization scripts were split into pre/post networking stages. They are now prefixed with loc- or net-. On a fresh install on the Ubuntu 16.04, the content of /etc/one-context.d should look like:

# ls /etc/one-context.d/
loc-05-grow-rootfs  loc-16-gen-env                net-15-hostname
loc-10-network      loc-20-set-username-password  net-97-start-script
loc-10-network-pci  loc-22-ssh_public_key         net-98-execute-scripts
loc-11-dns          loc-24-ssh_host_key           net-99-report-ready
loc-14-mount-swap   net-11-fix-loopback

Based on your logs, your directory contains also the old unprefixed scripts:

Sep 20 11:34:42 ubuntu one-contextd[1259]: Processing network scripts
Sep 20 11:34:42 ubuntu one-contextd[1261]: Script 05-grow-rootfs: Starting ...
Sep 20 11:34:42 ubuntu one-contextd[1305]: Script 05-grow-rootfs output: NOCHANGE: partition 1 is size 14677983. it cannot be grown
Sep 20 11:34:42 ubuntu one-contextd[1306]: Script 05-grow-rootfs: Finished with exit code 1
...
Sep 20 11:35:07 vm-2580 one-contextd[2321]: Script 99-report-ready: Starting ...
Sep 20 11:35:08 vm-2580 one-contextd[2332]: Script 99-report-ready: Finished with exit code 0
Sep 20 11:35:08 vm-2580 one-contextd[2334]: Script net-11-fix-loopback: Starting ...
Sep 20 11:35:08 vm-2580 one-contextd[2338]: Script net-11-fix-loopback: Finished with exit code 0
Sep 20 11:35:08 vm-2580 one-contextd[2340]: Script net-15-hostname: Starting ...

Unprefixed scripts (if found) are started in the post-networking stage. Also, the location of state files was moved into /var/run/one-context/, you'll find there the one_env generated by the current scripts. Please, clean your image from the old scripts, it should behave.

BTW if you simply upgraded the old package with a new package and the old (unprefixed) scripts left there, I see it as a packaging problem. I'll check if it happens. Thank you.

[tkald]

Cleaned up old install and reinstalled context 5.4.1 version.
I did miss state files location change - now in /var/run/one-context/ folder everything seems ok.

There're still some errors in the syslog about one-context-reconfigure-delayed service...
context5.4.1.log.txt

YES, I've had serveral older context package installations, always upgraded to newer package version.

[vholer]

Problem with leaving old scripts was caused by the packager, which automatically marked all files in /etc as configuration. I made a change in packaging, thank you.

And, about the problem with failing reconfiguration, it's a known issue. You don't have to be worried about this service if it fails only during the boot. There are multiple services, 2 for boot-only contextualization (one-context-local and one-context), 2 for later reconfiguration (on NIC hotplug and CONTEXT ISO change). Unfortunately, although the reconfiguration services have proper dependencies so they should run after boot-only services, when a machine is booting, NIC cold plugs trigger also the reconfiguration via udev and (don't know why) ignores the service dependencies. The error you see is a very last check in the script to fail if this happens.

[tkald]

Thank you!