Plan rejections occur when the scheduler work and the leader plan
applier disagree on the feasibility of a plan. This may happen for valid
reasons: since Nomad does parallel scheduling, it is expected that
different workers will have a different state when computing placements.
As the final plan reaches the leader plan applier, it may no longer be
valid due to a concurrent scheduling taking up intended resources. In
these situations the plan applier will notify the worker that the plan
was rejected and that they should refresh their state before trying
again.
In some rare and unexpected circumstances it has been observed that
workers will repeatedly submit the same plan, even if they are always
rejected.
While the root cause is still unknown this mitigation has been put in
place. The plan applier will now track the history of plan rejections
per client and include in the plan result a list of node IDs that should
be set as ineligible if the number of rejections in a given time window
crosses a certain threshold. The window size and threshold value can be
adjusted in the server configuration.
To avoid marking several nodes as ineligible at one, the operation is rate
limited to 5 nodes every 30min, with an initial burst of 10 operations.
Fixes#13505
This fixes#13505 by treating reserved_ports like we treat a lot of jobspec settings: merging settings from more global stanzas (client.reserved.reserved_ports) "down" into more specific stanzas (client.host_networks[].reserved_ports).
As discussed in #13505 there are other options, and since it's totally broken right now we have some flexibility:
Treat overlapping reserved_ports on addresses as invalid and refuse to start agents. However, I'm not sure there's a cohesive model we want to publish right now since so much 0.9-0.12 compat code still exists! We would have to explain to folks that if their -network-interface and host_network addresses overlapped, they could only specify reserved_ports in one place or the other?! It gets ugly.
Use the global client.reserved.reserved_ports value as the default and treat host_network[].reserverd_ports as overrides. My first suggestion in the issue, but @groggemans made me realize the addresses on the agent's interface (as configured by -network-interface) may overlap with host_networks, so you'd need to remove the global reserved_ports from addresses shared with a shared network?! This seemed really confusing and subtle for users to me.
So I think "merging down" creates the most expressive yet understandable approach. I've played around with it a bit, and it doesn't seem too surprising. The only frustrating part is how difficult it is to observe the available addresses and ports on a node! However that's a job for another PR.
In OSS we can upsert an allocation without worrying about whether that
alloc is in a namespace that actually exists, but in ENT that upsert
will add to the namespace's quotas. Ensure we're doing so in this
secure variables RPC test to fix the test breaking in the ENT repo.
* Failing test and TODO for wildcard
* Alias the namespace query parameter for Evals
* eval: fix list when using ACLs and * namespace
Apply the same verification process as in job, allocs and scaling
policy list endpoints to handle the eval list when using an ACL token
with limited namespace support but querying using the `*` wildcard
namespace.
* changelog: add entry for #13530
* ui: set namespace when querying eval
Evals have a unique UUID as ID, but when querying them the Nomad API
still expects a namespace query param, otherwise it assumes `default`.
Co-authored-by: Luiz Aoqui <luiz@hashicorp.com>
When the `Full` flag is passed for key rotation, we kick off a core
job to decrypt and re-encrypt all the secure variables so that they
use the new key.
* SV: CAS
* Implement Check and Set for Delete and Upsert
* Reading the conflict from the state store
* Update endpoint for new error text
* Updated HTTP api tests
* Conflicts to the HTTP api
* SV: structs: Update SV time to UnixNanos
* update mock to UnixNano; refactor
* SV: encrypter: quote KeyID in error
* SV: mock: add mock for namespace w/ SV
We need to track per-namespace storage usage for secure variables even
in Nomad OSS so that a cluster can be seamlessly upgraded from OSS to
ENT without having to re-calculate quota usage.
Provide a hook in the upsert RPC for enforcement of quotas in
ENT. This will be a no-op in Nomad OSS.
Extend the GC job to support periodic key rotation.
Update the GC process to safely support signed workload identity. We
can't GC any key used to sign a workload identity. Finding which key
was used to sign every allocation will be expensive, but there are not
that many keys. This lets us take a conservative approach: find the
oldest live allocation and ensure that we don't GC any key older than
that key.
The blocking query for `Keyring.List` appended the keys for each pass
through the blocking query to the response. This results in mulitple
copies of keys in the response. Overwrite the `reply.Keys` field on
each pass through the blocking query to ensure we only get the
expected page of responses.
Adds a new policy block inside namespaces to control access to secure
variables on the basis of path, with support for globbing.
Splits out VerifyClaim from ResolveClaim.
The ServiceRegistration RPC only needs to be able to verify that a
claim is valid for some allocation in the store; it doesn't care about
implicit policies or capabilities. Split this out to its own method on
the server so that the SecureVariables RPC can reuse it as a separate
step from resolving policies (see next commit).
Support implicit policies based on workload identity
The `Encrypt` method generates an appropriately-sized nonce and uses
that buffer as the prefix for the ciphertext. This keeps the
ciphertext and nonce together for decryption, and reuses the buffer as
much as possible without presenting the temptation to reuse the
cleartext buffer owned by the caller.
We include the key ID as the "additional data" field that's used as an
extra input to the authentication signature, to provide additional
protection that a ciphertext originated with that key.
Refactors the locking for the keyring so that the public methods are
generally (with one commented exception) responsible for taking the
lock and then inner methods are assumed locked.
* Add Path only index for SecureVariables
* Add GetSecureVariablesByPrefix; refactor tests
* Add search for SecureVariables
* Add prefix search for secure variables
This PR splits SecureVariable into SecureVariableDecrypted and
SecureVariableEncrypted in order to use the type system to help
verify that cleartext secret material is not committed to file.
* Make Encrypt function return KeyID
* Split SecureVariable
Co-authored-by: Tim Gross <tgross@hashicorp.com>
In order to support implicit ACL policies for tasks to get their own
secrets, each task would need to have its own ACL token. This would
add extra raft overhead as well as new garbage collection jobs for
cleaning up task-specific ACL tokens. Instead, Nomad will create a
workload Identity Claim for each task.
An Identity Claim is a JSON Web Token (JWT) signed by the server’s
private key and attached to an Allocation at the time a plan is
applied. The encoded JWT can be submitted as the X-Nomad-Token header
to replace ACL token secret IDs for the RPCs that support identity
claims.
Whenever a key is is added to a server’s keyring, it will use the key
as the seed for a Ed25519 public-private private keypair. That keypair
will be used for signing the JWT and for verifying the JWT.
This implementation is a ruthlessly minimal approach to support the
secure variables feature. When a JWT is verified, the allocation ID
will be checked against the Nomad state store, and non-existent or
terminal allocation IDs will cause the validation to be rejected. This
is sufficient to support the secure variables feature at launch
without requiring implementation of a background process to renew
soon-to-expire tokens.
Replication for the secure variables keyring. Because only key
metadata is stored in raft, we need to distribute key material
out-of-band from raft replication. A goroutine runs on each server and
watches for changes to the `RootKeyMeta`. When a new key is received,
attempt to fetch the key from the leader. If the leader doesn't have
the key (which may happen if a key is rotated right before a leader
transition), try to get the key from any peer.
After internal design review, we decided to remove exposing algorithm
choice to the end-user for the initial release. We'll solve nonce
rotation by forcing rotations automatically on key GC (in a core job,
not included in this changeset). Default to AES-256 GCM for the
following criteria:
* faster implementation when hardware acceleration is available
* FIPS compliant
* implementation in pure go
* post-quantum resistance
Also fixed a bug in the decoding from keystore and switched to a
harder-to-misuse encoding method.
The core jobs to garbage collect unused keys and perform full key
rotations will need to be able to query secure variables by key ID for
efficiency. Add an index to the state store and associated query
function and test.
When a server becomes leader, it will check if there are any keys in
the state store, and create one if there is not. The key metadata will
be replicated via raft to all followers, who will then get the key
material via key replication (not implemented in this changeset).
This changeset implements the keystore serialization/deserialization:
* Adds a JSON serialization extension for the `RootKey` struct, along with a metadata stub. When we serialize RootKey to the on-disk keystore, we want to base64 encode the key material but also exclude any frequently-changing fields which are stored in raft.
* Implements methods for loading/saving keys to the keystore.
* Implements methods for restoring the whole keystore from disk.
* Wires it all up with the `Keyring` RPC handlers and fixes up any fallout on tests.
Implement the basic upsert, list, and delete operations for
`RootKeyMeta` needed by the Keyring RPCs.
This changeset also implements two convenience methods
`RootKeyMetaByID` and `GetActiveRootKeyMeta` which are useful for
testing but also will be needed to implement the rest of the RPCs.
Stream snapshot to FSM when restoring from archive
The `RestoreFromArchive` helper decompresses the snapshot archive to a
temporary file before reading it into the FSM. For large snapshots
this performs a lot of disk IO. Stream decompress the snapshot as we
read it, without first writing to a temporary file.
Add bexpr filters to the `RestoreFromArchive` helper.
The operator can pass these as `-filter` arguments to `nomad operator
snapshot state` (and other commands in the future) to include only
desired data when reading the snapshot.
Whenever a node joins the cluster, either for the first time or after
being `down`, we emit a evaluation for every system job to ensure all
applicable system jobs are running on the node.
This patch adds an optimization to skip creating evaluations for system
jobs not in the current node's DC. While the scheduler performs the same
feasability check, skipping the creation of the evaluation altogether
saves disk, network, and memory.
This PR fixes a bug where client configuration max_kill_timeout was
not being enforced. The feature was introduced in 9f44780 but seems
to have been removed during the major drivers refactoring.
We can make sure the value is enforced by pluming it through the DriverHandler,
which now uses the lesser of the task.killTimeout or client.maxKillTimeout.
Also updates Event.SetKillTimeout to require both the task.killTimeout and
client.maxKillTimeout so that we don't make the mistake of using the wrong
value - as it was being given only the task.killTimeout before.
api: apply new ACL check for wildcard namespace
In #13606 the ACL check was refactored to better support the all
namespaces wildcard (`*`). This commit applies the changes to the jobs
and alloc list endpoints.
Improve how the all namespaces wildcard (`*`) is handled when checking
ACL permissions. When using the wildcard namespace the `AllowNsOp` would
return false since it looks for a namespace called `*` to match.
This commit changes this behavior to return `true` when the queried
namespace is `*` and the token allows the operation in _any_ namespace.
Actual permission must be checked per object. The helper function
`AllowNsOpFunc` returns a function that can be used to make this
verification.
* core: allow pause/un-pause of eval broker on region leader.
* agent: add ability to pause eval broker via scheduler config.
* cli: add operator scheduler commands to interact with config.
* api: add ability to pause eval broker via scheduler config
* e2e: add operator scheduler test for eval broker pause.
* docs: include new opertor scheduler CLI and pause eval API info.
It appears way back when this was first implemented in
9a917281af9c0a97a6c59575eaa52c5c86ffc60d, it was renamed from
NodeEvict (with a correct comment) to NodeUpdate. The comment was
changed from referring to only evictions to referring to "all allocs" in
the first sentence and "stop or evict" in the second.
This confuses every time I see it because I read the name (NodeUpdate)
and first sentence ("all the allocs") and assume this represents *all*
allocations... which isn't true.
I'm going to assume I'm the only one who doesn't read the 2nd sentence
and that's why this suboptimal wording has lasted 7 years, but can we
change it for my sake?
When calculating a services namespace for registration, the code
assumed the first task within the task array would include a
service block. This is incorrect as it is possible only a latter
task within the array contains a service definition.
This change fixes the logic, so we correctly search for a service
definition before identifying the namespace.
This PR adds the 'choose' query parameter to the '/v1/service/<service>' endpoint.
The value of 'choose' is in the form '<number>|<key>', number is the number
of desired services and key is a value unique but consistent to the requester
(e.g. allocID).
Folks aren't really expected to use this API directly, but rather through consul-template
which will soon be getting a new helper function making use of this query parameter.
Example,
curl 'localhost:4646/v1/service/redis?choose=2|abc123'
Note: consul-templte v0.29.1 includes the necessary nomadServices functionality.
The plan applier has to get a snapshot with a minimum index for the
plan it's working on in order to ensure consistency. Under heavy raft
loads, we can exceed the timeout. When this happens, we hit a bug
where the plan applier blocks waiting on the `indexCh` forever, and
all schedulers will block in `Plan.Submit`.
Closing the `indexCh` when the `asyncPlanWait` is done with it will
prevent the deadlock without impacting correctness of the previous
snapshot index.
This changeset includes the a PoC failing test that works by injecting
a large timeout into the state store. We need to turn this into a test
we can run normally without breaking the state store before we can
merge this PR.
Increase `snapshotMinIndex` timeout to 10s.
This timeout creates backpressure where any concurrent `Plan.Submit`
RPCs will block waiting for results. This sheds load across all
servers and gives raft some CPU to catch up, because schedulers won't
dequeue more work while waiting. Increase it to 10s based on
observations of large production clusters.
If the node has been GC'd or is down, we can't send it a node
unpublish. The CSI spec requires that we don't send the controller
unpublish before the node unpublish, but in the case where a node is
gone we can't know the final fate of the node unpublish step.
The `csi_hook` on the client will unpublish if the allocation has
stopped and if the host is terminated there's no mount for the volume
anyways. So we'll now assume that the node has unpublished at its
end. If it hasn't, any controller unpublish will potentially hang or
error and need to be retried.
When deleting evaluations and allocations during a reap event, the
index table entries for evals and allocs was updated irregardless
of whether changes were made.
This change modifies the state logic so that the index table is
only modified when the corresponding table has actually been
modified. Along with matching expected behaviour, this change has
the potential to reduce the number of times blocking queries will
return without any real state change.
