When a Nomad client that is running an allocation with
`max_client_disconnect` set misses a heartbeat the Nomad server will
update its status to `disconnected`.
Upon reconnecting, the client will make three main RPC calls:
- `Node.UpdateStatus` is used to set the client status to `ready`.
- `Node.UpdateAlloc` is used to update the client-side information about
allocations, such as their `ClientStatus`, task states etc.
- `Node.Register` is used to upsert the entire node information,
including its status.
These calls are made concurrently and are also running in parallel with
the scheduler. Depending on the order they run the scheduler may end up
with incomplete data when reconciling allocations.
For example, a client disconnects and its replacement allocation cannot
be placed anywhere else, so there's a pending eval waiting for
resources.
When this client comes back the order of events may be:
1. Client calls `Node.UpdateStatus` and is now `ready`.
2. Scheduler reconciles allocations and places the replacement alloc to
the client. The client is now assigned two allocations: the original
alloc that is still `unknown` and the replacement that is `pending`.
3. Client calls `Node.UpdateAlloc` and updates the original alloc to
`running`.
4. Scheduler notices too many allocs and stops the replacement.
This creates unnecessary placements or, in a different order of events,
may leave the job without any allocations running until the whole state
is updated and reconciled.
To avoid problems like this clients must update _all_ of its relevant
information before they can be considered `ready` and available for
scheduling.
To achieve this goal the RPC endpoints mentioned above have been
modified to enforce strict steps for nodes reconnecting:
- `Node.Register` does not set the client status anymore.
- `Node.UpdateStatus` sets the reconnecting client to the `initializing`
status until it successfully calls `Node.UpdateAlloc`.
These changes are done server-side to avoid the need of additional
coordination between clients and servers. Clients are kept oblivious of
these changes and will keep making these calls as they normally would.
The verification of whether allocations have been updates is done by
storing and comparing the Raft index of the last time the client missed
a heartbeat and the last time it updated its allocations.
This changeset allows Workload Identities to authenticate to all the RPCs that
support HTTP API endpoints, for use with PR #15864.
* Extends the work done for pre-forwarding authentication to all RPCs that
support a HTTP API endpoint.
* Consolidates the auth helpers used by the CSI, Service Registration, and Node
endpoints that are currently used to support both tokens and client secrets.
Intentionally excluded from this changeset:
* The Variables endpoint still has custom handling because of the implicit
policies. Ideally we'll figure out an efficient way to resolve those into real
policies and then we can get rid of that custom handling.
* The RPCs that don't currently support auth tokens (i.e. those that don't
support HTTP endpoints) have not been updated with the new pre-forwarding auth
We'll be doing this under a separate PR to support RPC rate metrics.
If a consumer of the new `Authenticate` method gets passed a bogus token that's
a correctly-shaped UUID, it will correctly get an identity without a ACL
token. But most consumers will then panic when they consume this nil `ACLToken`
for authorization.
Because no API client should ever send a bogus auth token, update the
`Authenticate` method to create the identity with remote IP (for metrics
tracking) but also return an `ErrPermissionDenied`.
Implement a metric for RPC requests with labels on the identity, so that
administrators can monitor the source of requests within the cluster. This
changeset demonstrates the change with the new `ACL.WhoAmI` RPC, and we'll wire
up the remaining RPCs once we've threaded the new pre-forwarding authentication
through the all.
Note that metrics are measured after we forward but before we return any
authentication error. This ensures that we only emit metrics on the server that
actually serves the request. We'll perform rate limiting at the same place.
Includes telemetry configuration to omit identity labels.
In #15417 we added a new `Authenticate` method to the server that returns an
`AuthenticatedIdentity` struct. This changeset implements this method for a
small number of RPC endpoints that together represent all the various ways in
which RPCs are sent, so that we can validate that we're happy with this
approach.
Upon dequeuing an evaluation workers snapshot their state store at the
eval's wait index or later. This ensures we process an eval at a point
in time after it was created or updated. Processing an eval on an old
snapshot could cause any number of problems such as:
1. Since job registration atomically updates an eval and job in a single
raft entry, scheduling against indexes before that may not have the
eval's job or may have an older version.
2. The older the scheduler's snapshot, the higher the likelihood
something has changed in the cluster state which will cause the plan
applier to reject the scheduler's plan. This could waste work or
even cause eval's to be failed needlessly.
However, the workers run in parallel with a new server pulling the
cluster state from a peer. During this time, which may be many minutes
long, the state store is likely far behind the minimum index required
to process evaluations.
This PR addresses this by adding an additional long backoff period after
an eval is nacked. If the scheduler's indexes catches up within the
additional backoff, it will unblock early to dequeue the next eval.
When the server shuts down we'll get a `context.Canceled` error from the state
store method. We need to bubble this error up so that other callers can detect
it. Handle this case separately when waiting after dequeue so that we can warn
on shutdown instead of throwing an ambiguous error message with just the text
"canceled."
While there may be more precise ways to block scheduling until the
server catches up, this approach adds little risk and covers additional
cases where a server may be temporarily behind due to a spike in load or
a saturated network.
For testing, we make the `raftSyncLimit` into a parameter on the worker's `run` method
so that we can run backoff tests without waiting 30+ seconds. We haven't followed thru
and made all the worker globals into worker parameters, because there isn't much
use outside of testing, but we can consider that in the future.
Co-authored-by: Tim Gross <tgross@hashicorp.com>
This changeset fixes a long-standing point of confusion in metrics emitted by
the eval broker. The eval broker has a queue of "blocked" evals that are waiting
for an in-flight ("unacked") eval of the same job to be completed. But this
"blocked" state is not the same as the `blocked` status that we write to raft
and expose in the Nomad API to end users. There's a second metric
`nomad.blocked_eval.total_blocked` that refers to evaluations in that
state. This has caused ongoing confusion in major customer incidents and even in
our own documentation! (Fixed in this PR.)
There's little functional change in this PR aside from the name of the metric
emitted, but there's a bit refactoring to clean up the names in `eval_broker.go`
so that there aren't name collisions and multiple names for the same
state. Changes included are:
* Everything that was previously called "pending" referred to entities that were
associated witht he "ready" metric. These are all now called "ready" to match
the metric.
* Everything named "blocked" in `eval_broker.go` is now named "pending", except
for a couple of comments that actually refer to blocked RPCs.
* Added a note to the upgrade guide docs for 1.5.0.
* Fixed the scheduling performance metrics docs because the description for
`nomad.broker.total_blocked` was actually the description for
`nomad.blocked_eval.total_blocked`.
* Add config elements
* Wire in snapshot configuration to raft
* Add hot reload of raft config
* Add documentation for new raft settings
* Add changelog
The description for the `nomad.nomad.blocked_evals.total_blocked` states that this could include evals blocked due to reached quota limits, but the `total_quota_limit` mentions being exclusive to its own metric. I personally interpret `total_blocked` as encompassing any blocked evals for any reason, as written in the docs. Though someone will have to verify the validity of that statement and possibly rectify the other metric description.
Fixed a typo: `limtis` vs `limits`.
* consul: correctly understand missing consul checks as unhealthy
This PR fixes a bug where Nomad assumed any registered Checks would exist
in the service registration coming back from Consul. In some cases, the
Consul may be slow in processing the check registration, and the response
object would not contain checks. Nomad would then scan the empty response
looking for Checks with failing health status, finding none, and then
marking a task/alloc as healthy.
In reality, we must always use Nomad's view of what checks should exist as
the source of truth, and compare that with the response Consul gives us,
making sure they match, before scanning the Consul response for failing
check statuses.
Fixes#15536
* consul: minor CR refactor using maps not sets
* consul: observe transition from healthy to unhealthy checks
* consul: spell healthy correctly
To see why I think this is a good change lets look at why I am making it
My disk was full, which means GC was happening agressively. So by the
time I called the logging endpoint from the SDK, the logs were GC'd
The error I was getting before was:
```
invalid character 'i' in literal false (expecting 'l')
```
Now the error I get is:
```
failed to decode log endpoint response as JSON: "failed to list entries: open /tmp/nomad.data.4219353875/alloc/f11fee50-2b66-a7a2-d3ec-8442cb3d557a/alloc/logs: no such file or directory"
```
Still not super descriptive but much more debugable
The OIDC provider cache is used by the RPC handler as the OIDC
implementation keeps long lived processes running. These process
include connections to the remote OIDC provider.
The Callback server is used by the CLI and starts when the login
command is triggered. This callback server includes success HTML
which is displayed when the user successfully logs into the remote
OIDC provider.
This adds new OIDC endpoints on the RPC endpoint. These two RPCs
handle generating the OIDC provider URL and then completing the
login by exchanging the provider token with an internal Nomad
token.
The RPC endpoints both do double forwarding. The initial forward
is to ensure we are talking to the regional leader; the second
then takes into account whether the auth method generates local or
global tokens. If it creates global tokens, we must then forward
onto the federated regional leader.
Luiz Aoqui