1335543731
The `ephemeral_disk` block's `migrate` field allows for best-effort migration of the ephemeral disk data to new nodes. The documentation says the `migrate` field is only respected if `sticky=true`, but in fact if client ACLs are not set the data is migrated even if `sticky=false`. The existing behavior when client ACLs are disabled has existed since the early implementation, so "fixing" that case now would silently break backwards compatibility. Additionally, having `migrate` not imply `sticky` seems nonsensical: it suggests that if we place on a new node we migrate the data but if we place on the same node, we throw the data away! Update so that `migrate=true` implies `sticky=true` as follows: * The failure mode when client ACLs are enabled comes from the server not passing along a migration token. Update the server so that the server provides a migration token whenever `migrate=true` and not just when `sticky=true` too. * Update the scheduler so that `migrate` implies `sticky`. * Update the client so that we check for `migrate || sticky` where appropriate. * Refactor the E2E tests to move them off the old framework and make the intention of the test more clear. |
||
|---|---|---|
| .. | ||
| acl | ||
| affinities | ||
| alloc_logs | ||
| artifact | ||
| auth | ||
| bin | ||
| clientstate | ||
| connect | ||
| consul | ||
| consultemplate | ||
| csi | ||
| deployment | ||
| disconnectedclients | ||
| e2eutil | ||
| eval_priority | ||
| events | ||
| example | ||
| execagent | ||
| framework | ||
| isolation | ||
| lifecycle | ||
| metrics | ||
| namespaces | ||
| networking | ||
| nodedrain | ||
| nomadexec | ||
| operator_scheduler | ||
| overlap | ||
| oversubscription | ||
| parameterized | ||
| periodic | ||
| podman | ||
| quotas | ||
| remotetasks | ||
| rescheduling | ||
| scaling | ||
| scalingpolicies | ||
| scheduler_sysbatch | ||
| scheduler_system | ||
| servicediscovery | ||
| spread | ||
| taskevents | ||
| terraform | ||
| ui | ||
| upgrades | ||
| vaultcompat | ||
| vaultsecrets | ||
| volumes | ||
| workload_id | ||
| .gitignore | ||
| README.md | ||
| e2e_test.go | ||
README.md
End to End Tests
This package contains integration tests. Unlike tests alongside Nomad code,
these tests expect there to already be a functional Nomad cluster accessible
(either on localhost or via the NOMAD_ADDR env var).
See framework/doc.go for how to write tests.
The NOMAD_E2E=1 environment variable must be set for these tests to run.
Provisioning Test Infrastructure on AWS
The terraform/ folder has provisioning code to spin up a Nomad cluster on
AWS. You'll need both Terraform and AWS credentials to setup AWS instances on
which e2e tests will run. See the
README
for details. The number of servers and clients is configurable, as is the
specific build of Nomad to deploy and the configuration file for each client
and server.
Provisioning Local Clusters
To run tests against a local cluster, you'll need to make sure the following environment variables are set:
NOMAD_ADDRshould point to one of the Nomad serversCONSUL_HTTP_ADDRshould point to one of the Consul serversNOMAD_E2E=1
TODO: the scripts in ./bin currently work only with Terraform, it would be
nice for us to have a way to deploy Nomad to Vagrant or local clusters.
Running
After completing the provisioning step above, you can set the client
environment for NOMAD_ADDR and run the tests as shown below:
# from the ./e2e/terraform directory, set your client environment
# if you haven't already
$(terraform output environment)
cd ..
go test -v ./...
If you want to run a specific suite, you can specify the -suite flag as
shown below. Only the suite with a matching Framework.TestSuite.Component
will be run, and all others will be skipped.
go test -v -suite=Consul .
If you want to run a specific test, you'll need to regex-escape some of the test's name so that the test runner doesn't skip over framework struct method names in the full name of the tests:
go test -v . -run 'TestE2E/Consul/\*consul\.ScriptChecksE2ETest/TestGroup'
^ ^ ^ ^
| | | |
Component | | Test func
| |
Go Package Struct
We're also in the process of migrating to "stdlib-style" tests that
use the standard go testing package without a notion of "suite". You
can run these with -run regexes the same way you would any other go
test:
go test -v . -run TestExample/TestExample_Simple
I Want To...
...SSH Into One Of The Test Machines
You can use the Terraform output to find the IP address. The keys will
in the ./terraform/keys/ directory.
ssh -i keys/nomad-e2e-*.pem ubuntu@${EC2_IP_ADDR}
Run terraform output for IP addresses and details.
...Deploy a Cluster of Mixed Nomad Versions
The variables.tf file describes the nomad_version, and
nomad_local_binary variables that can be used for most circumstances. But if
you want to deploy mixed Nomad versions, you can provide a list of versions in
your terraform.tfvars file.
For example, if you want to provision 3 servers all using Nomad 0.12.1, and 2 Linux clients using 0.12.1 and 0.12.2, you can use the following variables:
# will be used for servers
nomad_version = "0.12.1"
# will override the nomad_version for Linux clients
nomad_version_client_linux = [
"0.12.1",
"0.12.2"
]
...Deploy Custom Configuration Files
Set the profile field to "custom" and put the configuration files in
./terraform/config/custom/ as described in the
README.
...Deploy More Than 4 Linux Clients
Use the "custom" profile as described above.
...Change the Nomad Version After Provisioning
You can update the nomad_version variable, or simply rebuild the binary you
have at the nomad_local_binary path so that Terraform picks up the
changes. Then run terraform plan/terraform apply again. This will update
Nomad in place, making the minimum amount of changes necessary.