Add new workloads #25106
Add new workloads #25106
Conversation
| @@ -103,6 +103,12 @@ variable "aws_kms_alias" { | |||
| # provide a list of builds to override the values of nomad_sha, nomad_version, | |||
| # or nomad_local_binary. Most of the time you can ignore these variables! | |||
|
|
|||
| variable "nomad_local_binary_server_unique" { | |||
| description = "A nomad local binary paths to deploy to servers, to override nomad_local_binary" | |||
There was a problem hiding this comment.
I'm not sure I understand what this variable is supposed to do (what's "unique" mean here?), and the description matches nomad_local_binary_server exactly. It seems like we're overriding nomad_local_binary_server but I can't tell why. Shouldn't Enos set nomad_local_binary_server to make sure we're using the Linux binaries for the server instead?
There was a problem hiding this comment.
because nomad_local_binary_server is an array and we would need to pass the same path 3 times, but enos does not accept functions, so it would have to come from the previous step as a list or we would have to add an extra module to build the list
There was a problem hiding this comment.
Ah, ok... honestly I'm not sure that array has ever been used to have different values, so I wonder if it would be worth refactoring to just have a single override. But we can do that in a separate PR if you want.
There was a problem hiding this comment.
Im guessing at some point someone wanted to test compatibility of different versions?
There was a problem hiding this comment.
Yeah. I think I wrote that? 😊 But we really got away from running E2E manually, so there's probably a bunch of tiny design decisions that could be improved there.
Juanadelacuesta
force-pushed
the
f-NET-11478-enos-windows
branch
from
7157cbd to
b98792f
Compare
Juanadelacuesta
force-pushed
the
f-NET-11478-enos-windows
branch
from
b98792f to
40f537c
Compare
Juanadelacuesta
force-pushed
the
f-NET-11478-enos-windows
branch
2 times, most recently
from
8bbcb8a to
c5ae959
Compare
Juanadelacuesta
force-pushed
the
f-NET-11478-enos-windows
branch
from
c5ae959 to
51c80fe
Compare
Juanadelacuesta
force-pushed
the
f-NET-11478-enos-windows
branch
from
84b6710 to
1ba199f
Compare
Juanadelacuesta
force-pushed
the
f-NET-11478-enos-windows
branch
from
1ba199f to
25cf0ba
Compare
| workloads = { | ||
| service_raw_exec = { job_spec = "jobs/raw-exec-service.nomad.hcl", alloc_count = 3, type = "service" } | ||
| service_docker = { job_spec = "jobs/docker-service.nomad.hcl", alloc_count = 3, type = "service" } | ||
| system_docker = { job_spec = "jobs/docker-system.nomad.hcl", alloc_count = 0, type = "system" } |
There was a problem hiding this comment.
The alloc count for system jobs doesn't really matter, its set to 0 to emphasis it
| output "new_allocs_count" { | ||
| description = "The number of allocs that should be running in the cluster" | ||
| value = local.system_job_count * chomp(enos_local_exec.get_nodes.stdout) + local.service_batch_allocs | ||
| } |
There was a problem hiding this comment.
What's this output for? It doesn't match allocs_count
There was a problem hiding this comment.
Im trying to make this module aware of existing allocs so it outputs all the running allocs, new and old, and the output can be directly used for a next step in enos, because it does not accept functions as step.variables, so this output helps me debug
There was a problem hiding this comment.
Makes sense! Let's update the descriptions to make that clear. Right now this is the same as allocs_count.
| nomad_local_binary = step.copy_initial_binary.binary_path[matrix.os] | ||
| nomad_local_binary_server = step.copy_initial_binary.binary_path[local.server_os] |
There was a problem hiding this comment.
We should rebase this PR on #25172 once that's been merged.
| config { | ||
| image = "alpine:latest" | ||
| command = "sh" | ||
| args = ["-c", "while true; do sleep 30000; done"] |
There was a problem hiding this comment.
Batch workloads are a little tricky -- we probably want to make sure that they can complete and not get rescheduled by client/server restarts, rather than having them wait forever. But that'll require some new assertion logic, so let's come back to that.
| image = "alpine:latest" | ||
| command = "sh" | ||
| args = ["-c", "while true; do sleep 30000; done"] |
There was a problem hiding this comment.
For service/system jobs, we should probably use a workload where we can assert more about it than "is the alloc running?". busybox httpd would let us run a network service, so that we're exercising things like restoring CNI. Ok to leave for this PR, but let's come back to this too.
Co-authored-by: Tim Gross <tgross@hashicorp.com>
Description
This PR adds new system and batch workloads to the upgrade tests.
Testing & Reproduction steps
Links
Contributor Checklist
changelog entry using the
make clcommand.ensure regressions will be caught.
and job configuration, please update the Nomad website documentation to reflect this. Refer to
the website README for docs guidelines. Please also consider whether the
change requires notes within the upgrade guide.
Reviewer Checklist
backporting document.
in the majority of situations. The main exceptions are long-lived feature branches or merges where
history should be preserved.
within the public repository.