If a Nomad job is started with a large number of instances (e.g. 4 billion),
then the Nomad servers that attempt to schedule it will run out of memory and
crash. While it's unlikely that anyone would intentionally schedule a job with 4
billion instances, we have occasionally run into issues with bugs in external
automation. For example, an automated deployment system running on a test
environment had an off-by-one error, and deployed a job with count = uint32(-1),
causing the Nomad servers for that environment to run out of memory and crash.
To prevent this, this PR introduces a job_max_count Nomad server configuration
parameter. job_max_count limits the number of allocs that may be created from a
job. The default value is 50000 - this is low enough that a job with the maximum
possible number of allocs will not require much memory on the server, but is
still much higher than the number of allocs in the largest Nomad job we have
ever run.
Expand on the documentation of allocation garbage collection:
* Explain that server-side GC of allocations is tied to the GC of the
evaluation that spawned the allocation.
* Explain that server-side GC of allocations will force them to be immediately
GC'd on the client regardless of the client-side configurations.
Ref: https://github.com/hashicorp/nomad/issues/26765
Co-authored-by: Aimee Ukasick <Aimee.Ukasick@ibm.com>
Co-authored-by: Daniel Bennett <dbennett@hashicorp.com>
* Move commands from docs to its own root-level directory
* temporarily use modified dev-portal branch with nomad ia changes
* explicitly clone nomad ia exp branch
* retrigger build, fixed dev-portal broken build
* architecture, concepts and get started individual pages
* fix get started section destinations
* reference section
* update repo comment in website-build.sh to show branch
* docs nav file update capitalization
* update capitalization to force deploy
* remove nomad-vs-kubernetes dir; move content to what is nomad pg
* job section
* Nomad operations category, deploy section
* operations category, govern section
* operations - manage
* operations/scale; concepts scheduling fix
* networking
* monitor
* secure section
* remote auth-methods folder and move up pages to sso; linkcheck
* Fix install2deploy redirects
* fix architecture redirects
* Job section: Add missing section index pages
* Add section index pages so breadcrumbs build correctly
* concepts/index fix front matter indentation
* move task driver plugin config to new deploy section
* Finish adding full URL to tutorials links in nav
* change SSO to Authentication in nav and file system
* Docs NomadIA: Move tutorials into NomadIA branch (#26132)
* Move governance and policy from tutorials to docs
* Move tutorials content to job-declare section
* run jobs section
* stateful workloads
* advanced job scheduling
* deploy section
* manage section
* monitor section
* secure/acl and secure/authorization
* fix example that contains an unseal key in real format
* remove images from sso-vault
* secure/traffic
* secure/workload-identities
* vault-acl change unseal key and root token in command output sample
* remove lines from sample output
* fix front matter
* move nomad pack tutorials to tools
* search/replace /nomad/tutorials links
* update acl overview with content from deleted architecture/acl
* fix spelling mistake
* linkcheck - fix broken links
* fix link to Nomad variables tutorial
* fix link to Prometheus tutorial
* move who uses Nomad to use cases page; move spec/config shortcuts
add dividers
* Move Consul out of Integrations; move namespaces to govern
* move integrations/vault to secure/vault; delete integrations
* move ref arch to docs; rename Deploy Nomad back to Install Nomad
* address feedback
* linkcheck fixes
* Fixed raw_exec redirect
* add info from /nomad/tutorials/manage-jobs/jobs
* update page content with newer tutorial
* link updates for architecture sub-folders
* Add redirects for removed section index pages. Fix links.
* fix broken links from linkcheck
* Revert to use dev-portal main branch instead of nomadIA branch
* build workaround: add intro-nav-data.json with single entry
* fix content-check error
* add intro directory to get around Vercel build error
* workound for emtpry directory
* remove mdx from /intro/ to fix content-check and git snafu
* Add intro index.mdx so Vercel build should work
---------
Co-authored-by: Tu Nguyen <im2nguyen@gmail.com>
The server startup could "hang" to the view of an operator if it
had a key that could not be decrypted or replicated loaded from
the FSM at startup.
In order to prevent this happening, the server startup function
will now use a timeout to wait for the encrypter to be ready. If
the timeout is reached, the error is sent back to the caller which
fails the CLI command. This bubbling of error message will also
flush to logs which will provide addition operator feedback.
The server only cares about keys loaded from the FSM snapshot and
trailing logs before the encrypter should be classed as ready. So
that the encrypter ready function does not get blocked by keys
added outside of the initial Raft load, we take a snapshot of the
decryption tasks as we enter the blocking call, and class these as
our barrier.
Our vocabulary around scheduler behaviors outside of the `reschedule` and
`migrate` blocks leaves room for confusion around whether the reschedule tracker
should be propagated between allocations. There are effectively five different
behaviors we need to cover:
* restart: when the tasks of an allocation fail and we try to restart the tasks
in place.
* reschedule: when the `restart` block runs out of attempts (or the allocation
fails before tasks even start), and we need to move
the allocation to another node to try again.
* migrate: when the user has asked to drain a node and we need to move the
allocations. These are not failures, so we don't want to propagate the
reschedule tracker.
* replacement: when a node is lost, we don't count that against the `reschedule`
tracker for the allocations on the node (it's not the allocation's "fault",
after all). We don't want to run the `migrate` machinery here here either, as we
can't contact the down node. To the scheduler, this is effectively the same as
if we bumped the `group.count`
* replacement for `disconnect.replace = true`: this is a replacement, but the
replacement is intended to be temporary, so we propagate the reschedule tracker.
Add a section to the `reschedule`, `migrate`, and `disconnect` blocks explaining
when each item applies. Update the use of the word "reschedule" in several
places where "replacement" is correct, and vice-versa.
Fixes: https://github.com/hashicorp/nomad/issues/24918
Co-authored-by: Aimee Ukasick <aimee.ukasick@hashicorp.com>
In order to help users understand multi-region federated
deployments, this change adds two new sections to the website.
The first expands the architecture page, so we can add further
detail over time with an initial federation page. The second adds
a federation operations page which goes into failure planning and
mitigation.
Co-authored-by: Aimee Ukasick <aimee.ukasick@hashicorp.com>
Co-authored-by: Michael Schurter <mschurter@hashicorp.com>
When a root key is rotated, the servers immediately start signing Workload
Identities with the new active key. But workloads may be using those WI tokens
to sign into external services, which may not have had time to fetch the new
public key and which might try to fetch new keys as needed.
Add support for prepublishing keys. Prepublished keys will be visible in the
JWKS endpoint but will not be used for signing or encryption until their
`PublishTime`. Update the periodic key rotation to prepublish keys at half the
`root_key_rotation_threshold` window, and promote prepublished keys to active
after the `PublishTime`.
This changeset also fixes two bugs in periodic root key rotation and garbage
collection, both of which can't be safely fixed without implementing
prepublishing:
* Periodic root key rotation would never happen because the default
`root_key_rotation_threshold` of 720h exceeds the 72h maximum window of the FSM
time table. We now compare the `CreateTime` against the wall clock time instead
of the time table. (We expect to remove the time table in future work, ref
https://github.com/hashicorp/nomad/issues/16359)
* Root key garbage collection could GC keys that were used to sign
identities. We now wait until `root_key_rotation_threshold` +
`root_key_gc_threshold` before GC'ing a key.
* When rekeying a root key, the core job did not mark the key as inactive after
the rekey was complete.
Ref: https://hashicorp.atlassian.net/browse/NET-10398
Ref: https://hashicorp.atlassian.net/browse/NET-10280
Fixes: https://github.com/hashicorp/nomad/issues/19669
Fixes: https://github.com/hashicorp/nomad/issues/23528
Fixes: https://github.com/hashicorp/nomad/issues/19368
Our documentation has a hidden assumption that users know that federation
replication requires ACLs to be enabled and bootstrapped. Add notes at some of
the places users are likely to look for it.
A separate follow-up PR to the federation tutorial should point to the ACL
multi-region tutorial as well.
Fixes: https://github.com/hashicorp/nomad/issues/20128
Added the [OIDC Discovery](https://openid.net/specs/openid-connect-discovery-1_0.html) `/.well-known/openid-configuration` endpoint to Nomad, but it is only enabled if the `server.oidc_issuer` parameter is set. Documented the parameter, but without a tutorial trying to actually _use_ this will be very hard.
I intentionally did *not* use https://github.com/hashicorp/cap for the OIDC configuration struct because it's built to be a *compliant* OIDC provider. Nomad is *not* trying to be compliant initially because compliance to the spec does not guarantee it will actually satisfy the requirements of third parties. I want to avoid the problem where in an attempt to be standards compliant we ship configuration parameters that lock us in to a certain behavior that we end up regretting. I want to add parameters and behaviors as there's a demonstrable need.
Users always have the escape hatch of providing their own OIDC configuration endpoint. Nomad just needs to know the Issuer so that the JWTs match the OIDC configuration. There's no reason the actual OIDC configuration JSON couldn't live in S3 and get served directly from there. Unlike JWKS the OIDC configuration should be static, or at least change very rarely.
This PR is just the endpoint extracted from #18535. The `RS256` algorithm still needs to be added in hopes of supporting third parties such as [AWS IAM OIDC Provider](https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_providers_create_oidc.html).
Co-authored-by: Luiz Aoqui <luiz@hashicorp.com>
* api: enable support for setting original source alongside job
This PR adds support for setting job source material along with
the registration of a job.
This includes a new HTTP endpoint and a new RPC endpoint for
making queries for the original source of a job. The
HTTP endpoint is /v1/job/<id>/submission?version=<version> and
the RPC method is Job.GetJobSubmission.
The job source (if submitted, and doing so is always optional), is
stored in the job_submission memdb table, separately from the
actual job. This way we do not incur overhead of reading the large
string field throughout normal job operations.
The server config now includes job_max_source_size for configuring
the maximum size the job source may be, before the server simply
drops the source material. This should help prevent Bad Things from
happening when huge jobs are submitted. If the value is set to 0,
all job source material will be dropped.
* api: avoid writing var content to disk for parsing
* api: move submission validation into RPC layer
* api: return an error if updating a job submission without namespace or job id
* api: be exact about the job index we associate a submission with (modify)
* api: reword api docs scheduling
* api: prune all but the last 6 job submissions
* api: protect against nil job submission in job validation
* api: set max job source size in test server
* api: fixups from pr
In #13374 we updated the commented-out `license_path` in the packaged example
configuration file to match the existing documentation. Although this config
value was commented-out, it was reported that changing the value was
confusing. Update the commented-out line to the previous value and update the
documented examples to match that. This matches most of the examples for
Consul/Vault licensing as well. I've double-checked the tutorials and it looks
like it'd been left on the previous value there, so no additional work to be
done.
Prior to 2409f72 the code compared the modification index of a job to itself. Afterwards, the code compared the creation index of the job to itself. In either case there should never be a case of re-parenting of allocs causing the evaluation to trivially always result in false, which leads to unreclaimable memory.
Prior to this change allocations and evaluations for batch jobs were never garbage collected until the batch job was explicitly stopped. The new `batch_eval_gc_threshold` server configuration controls how often they are collected. The default threshold is `24h`.
* Add config elements
* Wire in snapshot configuration to raft
* Add hot reload of raft config
* Add documentation for new raft settings
* Add changelog
When replication of a single key fails, the replication loop breaks early and
therefore keys that fall later in the sorting order will never get
replicated. This is particularly a problem for clusters impacted by the bug that
caused #14981 and that were later upgraded; the keys that were never replicated
can now never be replicated, and so we need to handle them safely.
Included in the replication fix:
* Refactor the replication loop so that each key replicated in a function call
that returns an error, to make the workflow more clear and reduce nesting. Log
the error and continue.
* Improve stability of keyring replication tests. We no longer block leadership
on initializing the keyring, so there's a race condition in the keyring tests
where we can test for the existence of the root key before the keyring has
been initialize. Change this to an "eventually" test.
But these fixes aren't enough to fix#14981 because they'll end up seeing an
error once a second complaining about the missing key, so we also need to fix
keyring GC so the keys can be removed from the state store. Now we'll store the
key ID used to sign a workload identity in the Allocation, and we'll index the
Allocation table on that so we can track whether any live Allocation was signed
with a particular key ID.
* docs: write a lot of words about heartbeats
Alternative to #14670
* Apply suggestions from code review
Co-authored-by: Tim Gross <tgross@hashicorp.com>
* use descriptive title for link
* rework example of high failover ttl
Co-authored-by: Tim Gross <tgross@hashicorp.com>
The ACL command docs are now found within a sub-dir like the
operator command docs. Updates to the ACL token commands to
accommodate token expiry have also been added.
The ACL API docs are now found within a sub-dir like the operator
API docs. The ACL docs now include the ACL roles endpoint as well
as updated ACL token endpoints for token expiration.
The configuration section is also updated to accommodate the new
ACL and server parameters for the new ACL features.
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.
* 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.
Nomad inherited protocol version numbering configuration from Consul and
Serf, but unlike those projects Nomad has never used it. Nomad's
`protocol_version` has always been `1`.
While the code is effectively unused and therefore poses no runtime
risks to leave, I felt like removing it was best because:
1. Nomad's RPC subsystem has been able to evolve extensively without
needing to increment the version number.
2. Nomad's HTTP API has evolved extensively without increment
`API{Major,Minor}Version`. If we want to version the HTTP API in the
future, I doubt this is the mechanism we would choose.
3. The presence of the `server.protocol_version` configuration
parameter is confusing since `server.raft_protocol` *is* an important
parameter for operators to consider. Even more confusing is that
there is a distinct Serf protocol version which is included in `nomad
server members` output under the heading `Protocol`. `raft_protocol`
is the *only* protocol version relevant to Nomad developers and
operators. The other protocol versions are either deadcode or have
never changed (Serf).
4. If we were to need to version the RPC, HTTP API, or Serf protocols, I
don't think these configuration parameters and variables are the best
choice. If we come to that point we should choose a versioning scheme
based on the use case and modern best practices -- not this 6+ year
old dead code.
During incident response, operators may find that automated processes
elsewhere in the organization can be generating new workloads on Nomad
clusters that are unable to handle the workload. This changeset adds a
field to the `SchedulerConfiguration` API that causes all job
registration calls to be rejected unless the request has a management
ACL token.
FailoverHeartbeatTTL is the amount of time to wait after a server leader failure
before considering reallocating client tasks. This TTL should be fairly long as
the new server leader needs to rebuild the entire heartbeat map for the
cluster. In deployments with a small number of machines, the default TTL (5m)
may be unnecessary long. Let's allow operators to configure this value in their
config files.
This PR implements a new "System Batch" scheduler type. Jobs can
make use of this new scheduler by setting their type to 'sysbatch'.
Like the name implies, sysbatch can be thought of as a hybrid between
system and batch jobs - it is for running short lived jobs intended to
run on every compatible node in the cluster.
As with batch jobs, sysbatch jobs can also be periodic and/or parameterized
dispatch jobs. A sysbatch job is considered complete when it has been run
on all compatible nodes until reaching a terminal state (success or failed
on retries).
Feasibility and preemption are governed the same as with system jobs. In
this PR, the update stanza is not yet supported. The update stanza is sill
limited in functionality for the underlying system scheduler, and is
not useful yet for sysbatch jobs. Further work in #4740 will improve
support for the update stanza and deployments.
Closes#2527
This PR introduces the /v1/search/fuzzy API endpoint, used for fuzzy
searching objects in Nomad. The fuzzy search endpoint routes requests
to the Nomad Server leader, which implements the Search.FuzzySearch RPC
method.
Requests to the fuzzy search API are based on the api.FuzzySearchRequest
object, e.g.
{
"Text": "ed",
"Context": "all"
}
Responses from the fuzzy search API are based on the api.FuzzySearchResponse
object, e.g.
{
"Index": 27,
"KnownLeader": true,
"LastContact": 0,
"Matches": {
"tasks": [
{
"ID": "redis",
"Scope": [
"default",
"example",
"cache"
]
}
],
"evals": [],
"deployment": [],
"volumes": [],
"scaling_policy": [],
"images": [
{
"ID": "redis:3.2",
"Scope": [
"default",
"example",
"cache",
"redis"
]
}
]
},
"Truncations": {
"volumes": false,
"scaling_policy": false,
"evals": false,
"deployment": false
}
}
The API is tunable using the new server.search stanza, e.g.
server {
search {
fuzzy_enabled = true
limit_query = 200
limit_results = 1000
min_term_length = 5
}
}
These values can be increased or decreased, so as to provide more
search results or to reduce load on the Nomad Server. The fuzzy search
API can be disabled entirely by setting `fuzzy_enabled` to `false`.
