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 So we'll now assume that the node has unpublished at its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.
Almost all GC jobs check the index of the objects being GC'd to see if
they're older than a configured threshold. This code was repeated six
times in `CoreScheduler` with only logging changes, so it seems safe
to extract it as its own method.
Fix numerous go-getter security issues:
- Add timeouts to http, git, and hg operations to prevent DoS
- Add size limit to http to prevent resource exhaustion
- Disable following symlinks in both artifacts and `job run`
- Stop performing initial HEAD request to avoid file corruption on
retries and DoS opportunities.
**Approach**
Since Nomad has no ability to differentiate a DoS-via-large-artifact vs
a legitimate workload, all of the new limits are configurable at the
client agent level.
The max size of HTTP downloads is also exposed as a node attribute so
that if some workloads have large artifacts they can specify a high
limit in their jobspecs.
In the future all of this plumbing could be extended to enable/disable
specific getters or artifact downloading entirely on a per-node basis.
* test: use `T.TempDir` to create temporary test directory
This commit replaces `ioutil.TempDir` with `t.TempDir` in tests. The
directory created by `t.TempDir` is automatically removed when the test
and all its subtests complete.
Prior to this commit, temporary directory created using `ioutil.TempDir`
needs to be removed manually by calling `os.RemoveAll`, which is omitted
in some tests. The error handling boilerplate e.g.
defer func() {
if err := os.RemoveAll(dir); err != nil {
t.Fatal(err)
}
}
is also tedious, but `t.TempDir` handles this for us nicely.
Reference: https://pkg.go.dev/testing#T.TempDir
Signed-off-by: Eng Zer Jun <engzerjun@gmail.com>
* test: fix TestLogmon_Start_restart on Windows
Signed-off-by: Eng Zer Jun <engzerjun@gmail.com>
* test: fix failing TestConsul_Integration
t.TempDir fails to perform the cleanup properly because the folder is
still in use
testing.go:967: TempDir RemoveAll cleanup: unlinkat /tmp/TestConsul_Integration2837567823/002/191a6f1a-5371-cf7c-da38-220fe85d10e5/web/secrets: device or resource busy
Signed-off-by: Eng Zer Jun <engzerjun@gmail.com>
In #12324 we made it so that plugins wait until the node drain is
complete, as we do for system jobs. But we neglected to mark the node
drain as complete once only plugins (or system jobs) remaining, which
means that the node drain is left in a draining state until the
`deadline` time expires. This was incorrectly documented as expected
behavior in #12324.
After a more detailed analysis of this feature, the approach taken in
PR #12449 was found to be not ideal due to poor UX (users are
responsible for setting the entity alias they would like to use) and
issues around jobs potentially masquerading itself as another Vault
entity.
This PR introduces the `address` field in the `service` block so that Nomad
or Consul services can be registered with a custom `.Address.` to advertise.
The address can be an IP address or domain name. If the `address` field is
set, the `service.address_mode` must be set in `auto` mode.
This PR updates the changelog, adds notes the 1.3 upgrade guide, and
updates the connect integration docs with documentation about the new
requirement on Consul ACL policies of Consul agent default anonymous ACL
tokens.
* Add os to NodeListStub struct.
Signed-off-by: Shishir Mahajan <smahajan@roblox.com>
* Add os as a query param to /v1/nodes.
Signed-off-by: Shishir Mahajan <smahajan@roblox.com>
* Add test: os as a query param to /v1/nodes.
Signed-off-by: Shishir Mahajan <smahajan@roblox.com>
The CSI HTTP API has to transform the CSI volume to redact secrets,
remove the claims fields, and to consolidate the allocation stubs into
a single slice of alloc stubs. This was done manually in #8590 but
this is a large amount of code and has proven both very bug prone
(see #8659, #8666, #8699, #8735, and #12150) and requires updating
lots of code every time we add a field to volumes or plugins.
In #10202 we introduce encoding improvements for the `Node` struct
that allow a more minimal transformation. Apply this same approach to
serializing `structs.CSIVolume` to API responses.
Also, the original reasoning behind #8590 for plugins no longer holds
because the counts are now denormalized within the state store, so we
can simply remove this transformation entirely.
* services: add pagination and filter support to info RPC.
* cli: add filter flag to service info command.
* docs: add pagination and filter details to services info API.
* paginator: minor updates to comment and func signature.
Many of our scripts have a non-portable interpreter line for bash and
use bash-specific variables like `BASH_SOURCE`. Update the interpreter
line to be portable between various Linuxes and macOS without
complaint from posix shell users.
* planner: expose ServerMeetsMinimumVersion via Planner interface
* filterByTainted: add flag indicating disconnect support
* allocReconciler: accept and pass disconnect support flag
* tests: update dependent tests
Move some common Vault API data struct decoding out of the Vault client
so it can be reused in other situations.
Make Vault job validation its own function so it's easier to expand it.
Rename the `Job.VaultPolicies` method to just `Job.Vault` since it
returns the full Vault block, not just their policies.
Set `ChangeMode` on `Vault.Canonicalize`.
Add some missing tests.
Allows specifying an entity alias that will be used by Nomad when
deriving the task Vault token.
An entity alias assigns an indentity to a token, allowing better control
and management of Vault clients since all tokens with the same indentity
alias will now be considered the same client. This helps track Nomad
activity in Vault's audit logs and better control over Vault billing.
Add support for a new Nomad server configuration to define a default
entity alias to be used when deriving Vault tokens. This default value
will be used if the task doesn't have an entity alias defined.
This PR adds support for the raw_exec driver on systems with only cgroups v2.
The raw exec driver is able to use cgroups to manage processes. This happens
only on Linux, when exec_driver is enabled, and the no_cgroups option is not
set. The driver uses the freezer controller to freeze processes of a task,
issue a sigkill, then unfreeze. Previously the implementation assumed cgroups
v1, and now it also supports cgroups v2.
There is a bit of refactoring in this PR, but the fundamental design remains
the same.
Closes#12351#12348
The volume watcher design was based on deploymentwatcher and drainer,
but has an important difference: we don't want to maintain a goroutine
for the lifetime of the volume. So we stop the volumewatcher goroutine
for a volume when that volume has no more claims to free. But the
shutdown races with updates on the parent goroutine, and it's possible
to drop updates. Fortunately these updates are picked up on the next
core GC job, but we're most likely to hit this race when we're
replacing an allocation and that's the time we least want to wait.
Wait until the volume has "settled" before stopping this goroutine so
that the race between shutdown and the parent goroutine sending on
`<-updateCh` is pushed to after the window we most care about quick
freeing of claims.
* Fixes a resource leak when volumewatchers are no longer needed. The
volume is nil and can't ever be started again, so the volume's
`watcher` should be removed from the top-level `Watcher`.
* De-flakes the GC job test: the test throws an error because the
claimed node doesn't exist and is unreachable. This flaked instead of
failed because we didn't correctly wait for the first pass through the
volumewatcher.
Make the GC job wait for the volumewatcher to reach the quiescent
timeout window state before running the GC eval under test, so that
we're sure the GC job's work isn't being picked up by processing one
of the earlier claims. Update the claims used so that we're sure the
GC pass won't hit a node unpublish error.
* Adds trace logging to unpublish operations
In the same manner as the delete RPC, the list and read service
registration endpoints can be called either by external operators
or Nomad nodes. The latter occurs when a template is being
rendered which includes Nomad API template funcs. In this case,
the auth token is looked up as the node secret ID for auth.
* lint: require should not be aliased in core_sched_test
* lint: require should not be aliased in volumes_watcher_test
* testing: don't alias state package in core_sched_test
In #12112 and #12113 we solved for the problem of races in releasing
volume claims, but there was a case that we missed. During a node
drain with a controller attach/detach, we can hit a race where we call
controller publish before the unpublish has completed. This is
discouraged in the spec but plugins are supposed to handle it
safely. But if the storage provider's API is slow enough and the
plugin doesn't handle the case safely, the volume can get "locked"
into a state where the provider's API won't detach it cleanly.
Check the claim before making any external controller publish RPC
calls so that Nomad is responsible for the canonical information about
whether a volume is currently claimed.
This has a couple side-effects that also had to get fixed here:
* Changing the order means that the volume will have a past claim
without a valid external node ID because it came from the client, and
this uncovered a separate bug where we didn't assert the external node
ID was valid before returning it. Fallthrough to getting the ID from
the plugins in the state store in this case. We avoided this
originally because of concerns around plugins getting lost during node
drain but now that we've fixed that we may want to revisit it in
future work.
* We should make sure we're handling `FailedPrecondition` cases from
the controller plugin the same way we handle other retryable cases.
* Several tests had to be updated because they were assuming we fail
in a particular order that we're no longer doing.
Revert a small part of #11600 after @lgfa29 discovered it would break
compatibility with Nomad <= v1.2!
Nomad <= v1.2 expects the `vsn` tag to exist in Serf. It has always been
`1`. It has no functional purpose. However it causes a parsing error if
it is not set:
https://github.com/hashicorp/nomad/blob/v1.2.6/nomad/util.go#L103-L108
This means Nomad servers at version v1.2 or older will not allow servers
without this tag to join.
The `mvn` minor version tag is also checked, but soft fails. I'm not
setting that because I want as much of this cruft gone as possible.
Downgrading the Raft version protocol is not a supported operation.
Checking for a downgrade is hard since this information is not stored in
any persistent place. When a server re-joins a cluster with a prior Raft
version, the Serf tag is updated so Nomad can't tell that the version
changed.
Mixed version clusters must be supported to allow for zero-downtime
rolling upgrades. During this it's expected that the cluster will have
mixed Raft versions. Enforcing consistency strong version consistency
would disrupt this flow.
The approach taken here is to store the Raft version on disk. When the
server starts the `raft_protocol` value is written to the file
`data_dir/raft/version`. If that file already exists, its content is
checked against the current `raft_protocol` value to detect downgrades
and prevent the server from starting.
Any other types of errors are ignore to prevent disruptions that are
outside the control of operators. The only option in cases of an invalid
or corrupt file would be to delete it, making this check useless. So
just overwrite its content with the new version and provide guidance on
how to check that their cluster is an expected state.
Pass-through the `-secret` and `-parameter` flags to allow setting
parameters for the snapshot and overriding the secrets we've stored on
the CSI volume in the state store.
Listing snapshots was incorrectly returning nanoseconds instead of
seconds, and formatting of timestamps both list and create snapshot
was treating the timestamp as though it were nanoseconds instead of
seconds. This resulted in create timestamps always being displayed as
zero values.
Fix the unit conversion error in the command line and the incorrect
extraction in the CSI plugin client code. Beef up the unit tests to
make sure this code is actually exercised.
A volume that has single-use access mode is feasibility checked during
scheduling to ensure that only a single reader or writer claim
exists. However, because feasibility checking is done one alloc at a
time before the plan is written, a job that's misconfigured to have
count > 1 that mounts one of these volumes will pass feasibility
checking.
Enforce the check at validation time instead to prevent us from even
trying to evaluation a job that's misconfigured this way.
When a node fails its heart beating a number of actions are taken
to ensure state is cleaned. Service registrations a loosely tied
to nodes, therefore we should remove these from state when a node
is considered terminally down.
When a node is garbage collected, we assume that the volume is no
longer attached to it and ignore the `ErrUnknownNode` error. But we
used `errors.Is` to check for a wrapped error, and RPC flattens the
errors during serialization. This results in an error check that works
in automated testing but not in real clusters. Use a string contains
check instead.
Raft v3 introduced a new API for adding and removing peers that takes
the peer ID instead of the address.
Prior to this change, Nomad would use the remote peer Raft version for
deciding which API to use, but this would not work in the scenario where
a Raft v3 server tries to remove a Raft v2 server; the code running uses
v3 so it's unable to call the v2 API.
This change uses the Raft version of the server running the code to
decide which API to use. If the remote peer is a Raft v2, it uses the
server address as the ID.
When a node is drained, system jobs are left until last so that
operators can rely on things like log shippers running even as their
applications are getting drained off. Include CSI plugins in this set
so that Controller plugins deployed as services can be handled as
gracefully as Node plugins that are running as system jobs.
* Fix plugin capability sorting.
The `sort.StringSlice` method in the stdlib doesn't actually sort, but
instead constructs a sorting type which you call `Sort()` on.
* Sort allocations for plugins by modify index.
Present allocations in modify index order so that newest allocations
show up at the top of the list. This results in sorted allocs in
`nomad plugin status :id`, just like `nomad job status :id`.
* Sort allocations for volumes in HTTP response.
Present allocations in modify index order so that newest allocations
show up at the top of the list. This results in sorted allocs in
`nomad volume status :id`, just like `nomad job status :id`.
This is implemented in the HTTP response and not in the state store
because the state store maintains two separate lists of allocs that
are merged before sending over the API.
* Fix length of alloc IDs in `nomad volume status` output
Part 2 of breaking up https://github.com/hashicorp/nomad/pull/12255
This PR makes it so gotestsum is invoked only in CircleCI. Also the
HCLogger(t) is plumbed more correctly in TestServer and TestAgent so
that they respect NOMAD_TEST_LOG_LEVEL.
The reason for these is we'll want to disable logging in GHA,
where spamming the disk with logs really drags performance.
The `related` query param is used to indicate that the request should
return a list of related (next, previous, and blocked) evaluations.
Co-authored-by: Jasmine Dahilig <jasmine@hashicorp.com>
When a Nomad server becomes the Raft leader, it must perform several
actions defined in the establishLeadership function. If any of these
actions fail, Raft will think the node is the leader, but it will not
actually be able to act as a Nomad leader.
In this scenario, leadership must be revoked and transferred to another
server if possible, or the node should retry the establishLeadership
steps.
The alloc list test with pagination was creating allocs before the
target namespace existed. This works in OSS but fails in ENT because
quotas are checked before the alloc can be created, so the namespace
must exist beforehand.
The `Job.List` RPC attaches a `JobSummary` to each job stub. We're
using the request namespace and not the job namespace for that query,
which results in a nil `JobSummary` whenever we pass the wildcard
namespace. This is incorrect and causes panics in downstream consumers
like the CLI, which assume the `JobSummary` is non-nil as an unstate
invariant.
CSI `CreateVolume` RPC is idempotent given that the topology,
capabilities, and parameters are unchanged. CSI volumes have many
user-defined fields that are immutable once set, and many fields that
are not user-settable.
Update the `Register` RPC so that updating a volume via the API merges
onto any existing volume without touching Nomad-controlled fields,
while validating it with the same strict requirements expected for
idempotent `CreateVolume` RPCs.
Also, clarify that this state store method is used for everything, not just
for the `Register` RPC.
The `CreateSnapshot` RPC expects a plugin ID to be set by the API, but
in the common case of the `nomad volume snapshot create` command, we
don't ask the user for the plugin ID because it's available from the
volume we're snapshotting.
Change the order of the RPC so that we get the volume first and then
use the volume's plugin ID for the plugin if the API didn't set the
value.
The `CSIPlugin.List` RPC was intended to accept a prefix to filter the
list of plugins being listed. This was being accidentally being done
in the state store instead, which contributed to incorrect filtering
behavior for plugins in the `volume plugin status` command.
Move the prefix matching into the RPC so that it calls the
prefix-matching method in the state store if we're looking for a
prefix.
Update the `plugin status command` to accept a prefix for the plugin
ID argument so that it matches the expected behavior of other commands.
When using a prefix value and the * wildcard for namespace, the endpoint
would not take the prefix value into consideration due to the order in
which the checks were executed but also the logic for retrieving volumes
from the state store.
This commit changes the order to check for a prefix first and wraps the
result iterator of the state store query in a filter to apply the
prefix.
The paginator logic was built when go-memdb iterators would return items
ordered lexicographically by their ID prefixes, but #12054 added the
option for some tables to return results ordered by their `CreateIndex`
instead, which invalidated the previous paginator assumption.
The iterator used for pagination must still return results in some order
so that the paginator can properly handle requests where the next_token
value is not present in the results anymore (e.g., the eval was GC'ed).
In these situations, the paginator will start the returned page in the
first element right after where the requested token should've been.
This commit moves the logic to generate pagination tokens from the
elements being paginated to the iterator itself so that callers can have
more control over the token format to make sure they are properly
ordered and stable.
It also allows configuring the paginator as being ordered in ascending
or descending order, which is relevant when looking for a token that may
not be present anymore.
This PR
- upgrades the serf library
- has the test start the join process using the un-joined server first
- disables schedulers on the servers
- uses the WaitForLeader and wantPeers helpers
Not sure which, if any of these actually improves the flakiness of this test.
* Remove redundant schedulable check in `FreeWriteClaims`. If a volume
has been created but not yet claimed, its capabilities will be checked
in `WriteSchedulable` at both scheduling time and claim time. We don't
need to also check them in the `FreeWriteClaims` method.
* Enforce maximum volume claims for writers.
When the scheduler checks feasibility for CSI volumes, the check is
fairly loose: earlier versions of the same job are not counted as
active claims. This allows the scheduler to place new allocations
for the new version of a job, under the assumption that we'll replace
the existing allocations and their volume claims.
But when the alloc runner claims the volume, we need to enforce the
active claims even if they're for allocations of an earlier version of
the job. Otherwise we'll try to mount a volume that's currently being
unmounted, and this will cause replacement allocations to frequently
fail.
* Enforce single-node reader check for read-only volumes. When the
alloc runner makes a claim for a read-only volume, we only check that
the volume is potentially schedulable and not that it actually has
free read claims.
If a plugin job fails before successfully fingerprinting the plugins,
the plugin will not exist when we try to delete the job. Tolerate
missing plugins.
The volumewatcher test incorrectly represents the change in attachment
and access modes introduced in Nomad 1.1.0 to support volume
creation. This leads to a test that happens to pass but only
accidentally.
Update the test to correctly represent the volume modes set by the
existing claims on the test volumes.
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.
These API endpoints now return results in chronological order. They
can return results in reverse chronological order by setting the
query parameter ascending=true.
- Eval.List
- Deployment.List
When an allocation is updated, the job summary for the associated job
is also updated. CSI uses the job summary to set the expected count
for controller and node plugins. We incorrectly used the allocation's
server status instead of the job status when deciding whether to
update or remove the job from the plugins. This caused a node drain or
other terminal state for an allocation to clear the expected count for
the entire plugin.
Use the job status to guide whether to update or remove the expected
count.
The existing CSI tests for the state store incorrectly modeled the
updates we received from servers vs those we received from clients,
leading to test assertions that passed when they should not.
Rework the tests to clarify each step in the lifecycle and rename CSI state
store functions for clarity
PR #11956 implemented a new mTLS RPC check to validate the role of the
certificate used in the request, but further testing revealed two flaws:
1. client-only endpoints did not accept server certificates so the
request would fail when forwarded from one server to another.
2. the certificate was being checked after the request was forwarded,
so the check would happen over the server certificate, not the
actual source.
This commit checks for the desired mTLS level, where the client level
accepts both, a server or a client certificate. It also validates the
cercertificate before the request is forwarded.
Non-CSI garbage collection tasks on the server only log the cutoff
index in the case where it's not a forced GC from `nomad system gc`.
Do the same for CSI for consistency.
Update the logic in the Nomad client's alloc health tracker which
erroneously marks existing healthy allocations with dead poststart ephemeral
tasks as unhealthy even if they were already successful during a previous
deployment.
This PR replaces use of time.After with a safe helper function
that creates a time.Timer to use instead. The new function returns
both a time.Timer and a Stop function that the caller must handle.
Unlike time.NewTimer, the helper function does not panic if the duration
set is <= 0.
The Plan.Submit endpoint assumed PlanRequest.Plan was never nil. While
there is no evidence it ever has been nil, we should not panic if a nil
plan is ever submitted because that would crash the leader.
* The volume claim GC method and volumewatcher both have logic
collecting terminal allocations that duplicates most of the logic
that's now in the state store's `CSIVolumeDenormalize` method. Copy
this logic into the state store so that all code paths have the same
view of the past claims.
* Remove logic in the volume claim GC that now lives in the state
store's `CSIVolumeDenormalize` method.
* Remove logic in the volumewatcher that now lives in the state
store's `CSIVolumeDenormalize` method.
* Remove logic in the node unpublish RPC that now lives in the state
store's `CSIVolumeDenormalize` method.
In the client's `(*csiHook) Postrun()` method, we make an unpublish
RPC that includes a claim in the `CSIVolumeClaimStateUnpublishing`
state and using the mode from the client. But then in the
`(*CSIVolume) Unpublish` RPC handler, we query the volume from the
state store (because we only get an ID from the client). And when we
make the client RPC for the node unpublish step, we use the _current
volume's_ view of the mode. If the volume's mode has been changed
before the old allocations can have their claims released, then we end
up making a CSI RPC that will never succeed.
Why does this code path get the mode from the volume and not the
claim? Because the claim written by the GC job in `(*CoreScheduler)
csiVolumeClaimGC` doesn't have a mode. Instead it just writes a claim
in the unpublishing state to ensure the volumewatcher detects a "past
claim" change and reaps all the claims on the volumes.
Fix this by ensuring that the `CSIVolumeDenormalize` creates past
claims for all nil allocations with a correct access mode set.
* csi: resolve invalid claim states on read
It's currently possible for CSI volumes to be claimed by allocations
that no longer exist. This changeset asserts a reasonable state at
the state store level by registering these nil allocations as "past
claims" on any read. This will cause any pass through the periodic GC
or volumewatcher to trigger the unpublishing workflow for those claims.
* csi: make feasibility check errors more understandable
When the feasibility checker finds we have no free write claims, it
checks to see if any of those claims are for the job we're currently
scheduling (so that earlier versions of a job can't block claims for
new versions) and reports a conflict if the volume can't be scheduled
so that the user can fix their claims. But when the checker hits a
claim that has a GCd allocation, the state is recoverable by the
server once claim reaping completes and no user intervention is
required; the blocked eval should complete. Differentiate the
scheduler error produced by these two conditions.
The volumewatcher that runs on the leader needs to make RPC calls
rather than writing to raft (as we do in the deploymentwatcher)
because the unpublish workflow needs to make RPC calls to the
clients. This requires that the volumewatcher has access to the
leader's ACL token.
But when leadership transitions, the new leader creates a new leader
ACL token. This ACL token needs to be passed into the volumewatcher
when we enable it, otherwise the volumewatcher can find itself with a
stale token.
* The volume claim GC method and volumewatcher both have logic
collecting terminal allocations that duplicates most of the logic
that's now in the state store's `CSIVolumeDenormalize` method. Copy
this logic into the state store so that all code paths have the same
view of the past claims.
* Remove logic in the volume claim GC that now lives in the state
store's `CSIVolumeDenormalize` method.
* Remove logic in the volumewatcher that now lives in the state
store's `CSIVolumeDenormalize` method.
* Remove logic in the node unpublish RPC that now lives in the state
store's `CSIVolumeDenormalize` method.
In the client's `(*csiHook) Postrun()` method, we make an unpublish
RPC that includes a claim in the `CSIVolumeClaimStateUnpublishing`
state and using the mode from the client. But then in the
`(*CSIVolume) Unpublish` RPC handler, we query the volume from the
state store (because we only get an ID from the client). And when we
make the client RPC for the node unpublish step, we use the _current
volume's_ view of the mode. If the volume's mode has been changed
before the old allocations can have their claims released, then we end
up making a CSI RPC that will never succeed.
Why does this code path get the mode from the volume and not the
claim? Because the claim written by the GC job in `(*CoreScheduler)
csiVolumeClaimGC` doesn't have a mode. Instead it just writes a claim
in the unpublishing state to ensure the volumewatcher detects a "past
claim" change and reaps all the claims on the volumes.
Fix this by ensuring that the `CSIVolumeDenormalize` creates past
claims for all nil allocations with a correct access mode set.
* csi: resolve invalid claim states on read
It's currently possible for CSI volumes to be claimed by allocations
that no longer exist. This changeset asserts a reasonable state at
the state store level by registering these nil allocations as "past
claims" on any read. This will cause any pass through the periodic GC
or volumewatcher to trigger the unpublishing workflow for those claims.
* csi: make feasibility check errors more understandable
When the feasibility checker finds we have no free write claims, it
checks to see if any of those claims are for the job we're currently
scheduling (so that earlier versions of a job can't block claims for
new versions) and reports a conflict if the volume can't be scheduled
so that the user can fix their claims. But when the checker hits a
claim that has a GCd allocation, the state is recoverable by the
server once claim reaping completes and no user intervention is
required; the blocked eval should complete. Differentiate the
scheduler error produced by these two conditions.
The volumewatcher that runs on the leader needs to make RPC calls
rather than writing to raft (as we do in the deploymentwatcher)
because the unpublish workflow needs to make RPC calls to the
clients. This requires that the volumewatcher has access to the
leader's ACL token.
But when leadership transitions, the new leader creates a new leader
ACL token. This ACL token needs to be passed into the volumewatcher
when we enable it, otherwise the volumewatcher can find itself with a
stale token.
The command line client sends a specific volume ID, but this isn't
enforced at the API level and we were incorrectly using a prefix match
for volume deregistration, resulting in cases where a volume with a
shorter ID that's a prefix of another volume would be deregistered
instead of the intended volume.
This PR exposes the following existing`consul-template` configuration options to Nomad jobspec authors in the `{job.group.task.template}` stanza.
- `wait`
It also exposes the following`consul-template` configuration to Nomad operators in the `{client.template}` stanza.
- `max_stale`
- `block_query_wait`
- `consul_retry`
- `vault_retry`
- `wait`
Finally, it adds the following new Nomad-specific configuration to the `{client.template}` stanza that allows Operators to set bounds on what `jobspec` authors configure.
- `wait_bounds`
Co-authored-by: Tim Gross <tgross@hashicorp.com>
Co-authored-by: Michael Schurter <mschurter@hashicorp.com>
## Development Environment Changes
* Added stringer to build deps
## New HTTP APIs
* Added scheduler worker config API
* Added scheduler worker info API
## New Internals
* (Scheduler)Worker API refactor—Start(), Stop(), Pause(), Resume()
* Update shutdown to use context
* Add mutex for contended server data
- `workerLock` for the `workers` slice
- `workerConfigLock` for the `Server.Config.NumSchedulers` and
`Server.Config.EnabledSchedulers` values
## Other
* Adding docs for scheduler worker api
* Add changelog message
Co-authored-by: Derek Strickland <1111455+DerekStrickland@users.noreply.github.com>
When `volumewatcher.Watcher` starts on the leader, it starts a watch
on every volume and triggers a reap of unused claims on any change to
that volume. But if a reaping is in-flight during leadership
transitions, it will fail and the event that triggered the reap will
be dropped. Perform one reap of unused claims at the start of the
watcher so that leadership transitions don't drop this event.
Some operators use very long group/task `shutdown_delay` settings to
safely drain network connections to their workloads after service
deregistration. But during incident response, they may want to cause
that drain to be skipped so they can quickly shed load.
Provide a `-no-shutdown-delay` flag on the `nomad alloc stop` and
`nomad job stop` commands that bypasses the delay. This sets a new
desired transition state on the affected allocations that the
allocation/task runner will identify during pre-kill on the client.
Note (as documented here) that using this flag will almost always
result in failed inbound network connections for workloads as the
tasks will exit before clients receive updated service discovery
information and won't be gracefully drained.
API queries can request pagination using the `NextToken` and `PerPage`
fields of `QueryOptions`, when supported by the underlying API.
Add a `NextToken` field to the `structs.QueryMeta` so that we have a
common field across RPCs to tell the caller where to resume paging
from on their next API call. Include this field on the `api.QueryMeta`
as well so that it's available for future versions of List HTTP APIs
that wrap the response with `QueryMeta` rather than returning a simple
list of structs. In the meantime callers can get the `X-Nomad-NextToken`.
Add pagination to the `Eval.List` RPC by checking for pagination token
and page size in `QueryOptions`. This will allow resuming from the
last ID seen so long as the query parameters and the state store
itself are unchanged between requests.
Add filtering by job ID or evaluation status over the results we get
out of the state store.
Parse the query parameters of the `Eval.List` API into the arguments
expected for filtering in the RPC call.
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.
Give ourselves some room for extension in the UI configuration block
by naming the field `ui_url`, which will let us have an `api_url`.
Fix the template path to ensure we're getting the right value from the
API.
When GetPolicy is called within the scaling handler, the index
table was being used to populate the reply index irregardless of
whether the policy was found or not. This change fixes that
behaviour so that the policy modify index is used when the policy
lookup is successful.
This change modifies the Nomad job register and deregister RPCs to
accept an updated option set which includes eval priority. This
param is optional and override the use of the job priority to set
the eval priority.
In order to ensure all evaluations as a result of the request use
the same eval priority, the priority is shared to the
allocReconciler and deploymentWatcher. This creates a new
distinction between eval priority and job priority.
The Nomad agent HTTP API has been modified to allow setting the
eval priority on job update and delete. To keep consistency with
the current v1 API, job update accepts this as a payload param;
job delete accepts this as a query param.
Any user supplied value is validated within the agent HTTP handler
removing the need to pass invalid requests to the server.
The register and deregister opts functions now all for setting
the eval priority on requests.
The change includes a small change to the DeregisterOpts function
which handles nil opts. This brings the function inline with the
RegisterOpts.
* api: return 404 for alloc FS list/stat endpoints
If the alloc filesystem doesn't have a file requested by the List
Files or Stat File API, we currently return a HTTP 500 error with the
expected "file not found" error message. Return a HTTP 404 error
instead.
* update FS Handler
Previously the FS handler would interpret a 500 status as a 404
in the adapter layer by checking if the response body contained
the text or is the response status
was 500 and then throw an error code for 404.
Co-authored-by: Jai Bhagat <jaybhagat841@gmail.com>
Enhance the CLI in order to return the host network in two flavors
(default, verbose) of the `node status` command.
Fixes: #11223.
Signed-off-by: Alessandro De Blasis <alex@deblasis.net>
Previously when creating an eval for job deregistration, the eval
priority was set to the default value irregardless of the job
priority. In situations where an operator would want to deregister
a high priority job so they could re-register; the evaluation may
get blocked for some time on a busy cluster because of the
deregsiter priority.
If a job had a lower than default priority and was deregistered,
the deregister eval would get a priority higher than that of the
job. If we attempted to register another job with a higher
priority than this, but still below the default, the deregister
would be actioned before the register.
Both situations described above seem incorrect and unexpected from
a user prespective.
This fix modifies to behaviour to set the deregister eval priority
to that of the job, if available. Otherwise the default value is
still used.
As we have continued to see reports of #9506 we need to elevate this log
line as it is the only way to detect when plans are being *erroneously*
rejected.
Users who see this log line repeatedly should drain and restart the node
in the log line. This seems to workaorund the issue.
Please post any details on #9506!
Fix a bug where the scheduler may panic when preemption is enabled. The conditions are a bit complicated:
A job with higher priority that schedule multiple allocations that preempt other multiple allocations on the same node, due to port/network/device assignments.
The cause of the bug is incidental mutation of internal cached data. `RankedNode` computes and cache proposed allocations in https://github.com/hashicorp/nomad/blob/v1.1.6/scheduler/rank.go#L42-L53 . But scheduler then mutates the list to remove pre-emptable allocs in https://github.com/hashicorp/nomad/blob/v1.1.6/scheduler/rank.go#L293-L294, and `RemoveAllocs` mutates and sets the tail of cached slice with `nil`s triggering a nil-pointer derefencing case.
I fixed the issue by avoiding the mutation in `RemoveAllocs` - the micro-optimization there doesn't seem necessary.
Fixes https://github.com/hashicorp/nomad/issues/11342
* Include region and namespace in CLI output
* Add region and prefix matching for server members
* Add namespace and region API outputs to cluster metadata folder
* Add region awareness to WaitForClient helper function
* Add helper functions for SliceStringHasPrefix and StringHasPrefixInSlice
* Refactor test client agent generation
* Add tests for region
* Add changelog
The fix seems to be related to the pointer comparison and swapping we
did around killing a non-leader. I actually can't quite explain it, but
when comparing against Consul's version of this test I noticed they used
the slice index to track the killed server instead of pointer swapping.
As soon as I switched to slice index tracking I could no longer
reproduce the failure.
In addition:
- Tested membership counts on all servers instead of just 1 for added
correctness.
- Stopped testing raft v1 because it is unsupported.
Add a new hostname string parameter to the network block which
allows operators to specify the hostname of the network namespace.
Changing this causes a destructive update to the allocation and it
is omitted if empty from API responses. This parameter also supports
interpolation.
In order to have a hostname passed as a configuration param when
creating an allocation network, the CreateNetwork func of the
DriverNetworkManager interface needs to be updated. In order to
minimize the disruption of future changes, rather than add another
string func arg, the function now accepts a request struct along with
the allocID param. The struct has the hostname as a field.
The in-tree implementations of DriverNetworkManager.CreateNetwork
have been modified to account for the function signature change.
In updating for the change, the enhancement of adding hostnames to
network namespaces has also been added to the Docker driver, whilst
the default Linux manager does not current implement it.
This fixes a bug in the event stream API where it currently interprets
namespace=* as an actual namespace, not a wildcard. When Nomad parses
incoming requests, it sets namespace to default if not specified, which
means the request namespace will never be an empty string, which is what
the event subscription was checking for. This changes the conditional
logic to check for a wildcard namespace instead of an empty one.
It also updates some event tests to include the default namespace in the
subscription to match current behavior.
Fixes#10903
When mTLS is enabled, only nomad servers of the region should access the
Raft RPC layer. Clients and servers in other regions should only use the
Nomad RPC endpoints.
Co-authored-by: Michael Schurter <mschurter@hashicorp.com>
Co-authored-by: Seth Hoenig <shoenig@hashicorp.com>
Attempt to deflake the test by avoiding shutting down the leaders, as leadership
recovery takes more time, and consequently longer to process raft configuration
changes and potentially failing the test.
When a node becomes ready, create an eval for all system jobs across
namespaces.
The previous code uses `job.ID` to deduplicate evals, but that ignores
the job namespace. Thus if there are multiple jobs in different
namespaces sharing the same ID/Name, only one will be considered for
running in the new node. Thus, Nomad may skip running some system jobs
in that node.
Fix a bug where system jobs may fail to be placed on a node that
initially was not eligible for system job placement.
This changes causes the reschedule to re-evaluate the node if any
attribute used in feasibility checks changes.
Fixes https://github.com/hashicorp/nomad/issues/8448
In a multi-task-group job, treat 0 canary groups as auto-promote.
This change fixes an edge case where Nomad requires a manual promotion,
if the job had any group with canary=0 and rest of groups having
auto_promote set.
Co-authored-by: Michael Schurter <mschurter@hashicorp.com>
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
Fix a panic in handling one-time auth tokens, used to support `nomad ui
--authenticate`.
If the nomad leader is a 1.1.x with some servers running as 1.0.x, the
pre-1.1.0 servers risk crashing and the cluster may lose quorum. That
can happen when `nomad authenticate -ui` command is issued, or when the
leader scans for expired tokens every 10 minutes.
Fixed#10943 .
Basically the same as #10896 but with the Affinity struct.
Since we use reflect.DeepEquals for job comparison, there is
risk of false positives for changes due to a job struct with
memoized vs non-memoized strings.
Closes#10897
This PR causes Nomad to no longer memoize the String value of
a Constraint. The private memoized variable may or may not be
initialized at any given time, which means a reflect.DeepEqual
comparison between two jobs (e.g. during Plan) may return incorrect
results.
Fixes#10836
When a task group with `service` block(s) is validated, we validate that there
are no duplicates, but this validation doesn't have access to the task environment
because it hasn't been created yet. Services and checks with interpolation can
be flagged incorrectly as conflicting. Name conflicts in services are not
actually an error in Consul and users have reported wanting to use the same
service name for task groups differentiated by tags.
Tim Gross