Panic on upgrade from 1.22

[Martin Packman]

The long-hop upgrade jobs in CI that test if it's possible to upgrade from 1.22 to master, currently fail with a panic during the upgrade steps:

<http://reports.vapour.ws/releases/3021/job/aws-upgrade-22-trusty-amd64/attempt/102>

<http://data.vapour.ws/juju-ci/products/version-3021/aws-upgrade-22-trusty-amd64/build-102/machine-0.log>

2015-08-31 14:20:10 INFO juju.worker.upgrader error.go:32 upgraded from 1.22.6-trusty-amd64 to 1.26-alpha1-trusty-amd64 ("https://172.31.13.242:17070/environment/7e544f12-f3a6-4535-839b-e171af224825/tools/1.26-alpha1-trusty-amd64")
2015-08-31 14:20:10 ERROR juju.cmd supercommand.go:430 must restart: an agent upgrade is available
2015-08-31 14:20:10 INFO juju.cmd supercommand.go:37 running jujud [1.26-alpha1-trusty-amd64 gc]
...
2015-08-31 14:20:17 INFO juju.upgrade upgrade.go:133 running upgrade step: move lastlogin and last connection to their own collections
2015-08-31 14:20:17 DEBUG juju.state open.go:306 closed state without error
panic: runtime error: invalid memory address or nil pointer dereference
[signal 0xb code=0x1 addr=0x0 pc=0x629acf]

goroutine 126 [running]:
runtime.panic(0x1a34800, 0x45315c8)
 /usr/lib/go/src/pkg/runtime/panic.c:266 +0xb6
github.com/juju/juju/state.MigrateLastLoginAndLastConnection(0xc2102ac210, 0x0, 0x0)
 /workspace/juju-core-1.26-alpha1/src/github.com/juju/juju/state/upgrades.go:1604 +0x15df
...

So, seems to be caused by the same branch that caused bug 1489896, but in a new form:

<https://github.com/juju/juju/pull/3151>

[Canonical Juju QA Bot (juju-qa-bot)]

Juju-CI verified that this issue is Fix Released in juju-core master:
    http://reports.vapour.ws/releases/3036

[Cheryl Jennings (cherylj)]

Review request for 1.25 backport: http://reviews.vapour.ws/r/2827/

[Canonical Juju QA Bot (juju-qa-bot)]

Juju-CI verified that this issue is Fix Released in juju-core 1.25:
    http://reports.vapour.ws/releases/3145