- On-the-wire gossip messages are sorted case sensitive, but in memory they are sorted case-insensitive.
- Add test for case insensitive match on multiple names.
This also eliminates the redefinition of 'start_container' by some of
the tests, which causes failures when multiple tests are run with
`run_all.sh`.
Fixes#1294.
- Nameserver holds a sorted list of (hostname, peer, container id, ip) tuples.
- Entries are gossiped using topology gossip / broadcast - only new/updated entries
are propogated as expected.
- Entries for a given peer are removed when the connection to the peer is lost.
- Entries are tombstoned when containers die. Tombstones are kept for 10mins,
as they do not need to survive partition, only gossip propagation delay.
This gives more consistent test execution times, and also make image
fetch failures more apparent and decoupled from the tests.
We could *always* just run pull, but it's an expensive operation, so
we check for the presence of the image first. The downside is that we
could be running old images. That is not a problem on our CI
infrastructure though since it spins up fresh VMs per for every tun.
If the name cannot be resolved, getent fails, which in turn causes
assert_dns_record to just return with a non-zero status at that point,
which causes the whole test to just terminate with a non-zero status,
but no indication whatsoever what had gone wrong.
Pass `-e WEAVE_CIDR=` to your container to use IPAM. If IPAM is not
enabled on the weave router, the container will be booted without weave
networking.
I'm not crazy about adding *another* undocumented weave cli command
(ipam-cidr), but it saves the proxy messing around finding container
IPs.
Bug came up in a test where we are restarting and doing an exec checking
a network interface right away. weavewait is only added if the container
has a WEAVE_CIDR var set, the same as at container creation.
Bryan Boreham