Commit Graph

12 Commits

Author SHA1 Message Date
Tim Gross c320c1ba57
CSI: fix struct copying errors (#9239)
The CSIVolume struct "denormalizes" allocations when it's first queried from
the state store. The CSIVolumeByID method on the state store copies the volume
before denormalizing so that we don't end up with unexpected changes. The
copying has some subtle bugs that meant that Allocations (as well as
Topologies and MountOptions) were not getting copied when expected.

Also, ensure we never write allocations attached to volumes to the state store
during claims.
2020-11-18 10:59:25 -05:00
Tim Gross 3ceb5b36b1
csi: allow more than 1 writer claim for multi-writer mode (#9040)
Fixes a bug where CSI volumes with the `MULTI_NODE_MULTI_WRITER` access mode
were using the same logic as `MULTI_NODE_SINGLE_WRITER` to determine whether
the volume had writer claims available for scheduling.

Extends CSI claim endpoint test to exercise multi-reader and make sure `WriteFreeClaims`
is exercised for multi-writer in feasibility test.
2020-10-07 10:43:23 -04:00
Lang Martin 7d483f93c0
csi: plugins track jobs in addition to allocations, and use job information to set expected counts (#8699)
* nomad/structs/csi: add explicit job support
* nomad/state/state_store: capture job updates directly
* api/nodes: CSIInfo needs the AllocID
* command/agent/csi_endpoint: AllocID was missing
Co-authored-by: Tim Gross <tgross@hashicorp.com>
2020-08-27 17:20:00 -04:00
Tim Gross 52e805a6a6
csi: ensure Read/WriteAllocs aren't released early (#7841)
We should only remove the `ReadAllocs`/`WriteAllocs` values for a
volume after the claim has entered the "ready to free"
state. The volume will eventually be released as expected. But
querying the volume API will show the volume is released before the
controller unpublish has finished and this can cause a race with
starting new jobs.

Test updates are to cover cases where we're dropping claims but not
running through the whole reaping process.
2020-04-30 17:11:31 -04:00
Tim Gross 083b35d651
csi: checkpoint volume claim garbage collection (#7782)
Adds a `CSIVolumeClaim` type to be tracked as current and past claims
on a volume. Allows for a client RPC failure during node or controller
detachment without having to keep the allocation around after the
first garbage collection eval.

This changeset lays groundwork for moving the actual detachment RPCs
into a volume watching loop outside the GC eval.
2020-04-23 11:06:23 -04:00
Lang Martin ea80330aaa
csi: nomad/structs: test volume denormalize without plugin (#7472) 2020-03-26 09:43:59 -04:00
Lang Martin d994990ef0
csi: the scheduler allows a job with a volume write claim to be updated (#7438)
* nomad/structs/csi: split CanWrite into health, in use

* scheduler/scheduler: expose AllocByID in the state interface

* nomad/state/state_store_test

* scheduler/stack: SetJobID on the matcher

* scheduler/feasible: when a volume writer is in use, check if it's us

* scheduler/feasible: remove SetJob

* nomad/state/state_store: denormalize allocs before Claim

* nomad/structs/csi: return errors on claim, with context

* nomad/csi_endpoint_test: new alloc doesn't look like an update

* nomad/state/state_store_test: change test reference to CanWrite
2020-03-23 21:21:04 -04:00
Tim Gross 0cd2d3cc29 csi: make claims on volumes idempotent for the same alloc (#7328)
Nomad clients will push node updates during client restart which can
cause an extra claim for a volume by the same alloc. If an alloc
already claims a volume, we can allow it to be treated as a valid
claim and continue.
2020-03-23 13:58:30 -04:00
Lang Martin 369b0e54b9 csi: volumes use `Schedulable` rather than `Healthy` (#7250)
* structs: add ControllerRequired, volume.Name, no plug.Type

* structs: Healthy -> Schedulable

* state_store: Healthy -> Schedulable

* api: add ControllerRequired to api data types

* api: copy csi structs changes

* nomad/structs/csi: include name and external id

* api/csi: include Name and ExternalID

* nomad/structs/csi: comments for the 3 ids
2020-03-23 13:58:30 -04:00
Lang Martin 7b675f89ac csi: fix index maintenance for CSIVolume and CSIPlugin tables (#7049)
* state_store: csi volumes/plugins store the index in the txn

* nomad: csi_endpoint_test require index checks need uint64()

* nomad: other tests using int 0 not uint64(0)

* structs: pass index into New, but not other struct methods

* state_store: csi plugin indexes, use new struct interface

* nomad: csi_endpoint_test check index/query meta (on explicit 0)

* structs: NewCSIVolume takes an index arg now

* scheduler/test: NewCSIVolume takes an index arg now
2020-03-23 13:58:29 -04:00
Lang Martin 88316208a0 csi: server-side plugin state tracking and api (#6966)
* structs: CSIPlugin indexes jobs acting as plugins and node updates

* schema: csi_plugins table for CSIPlugin

* nomad: csi_endpoint use vol.Denormalize, plugin requests

* nomad: csi_volume_endpoint: rename to csi_endpoint

* agent: add CSI plugin endpoints

* state_store_test: use generated ids to avoid t.Parallel conflicts

* contributing: add note about registering new RPC structs

* command: agent http register plugin lists

* api: CSI plugin queries, ControllerHealthy -> ControllersHealthy

* state_store: copy on write for volumes and plugins

* structs: copy on write for volumes and plugins

* state_store: CSIVolumeByID returns an unhealthy volume, denormalize

* nomad: csi_endpoint use CSIVolumeDenormalizePlugins

* structs: remove struct errors for missing objects

* nomad: csi_endpoint return nil for missing objects, not errors

* api: return meta from Register to avoid EOF error

* state_store: CSIVolumeDenormalize keep allocs in their own maps

* state_store: CSIVolumeDeregister error on missing volume

* state_store: CSIVolumeRegister set indexes

* nomad: csi_endpoint use CSIVolumeDenormalizePlugins tests
2020-03-23 13:58:29 -04:00
Lang Martin e922531aaf structs: move the content of csi_volumes into csi 2020-03-23 13:58:29 -04:00
Renamed from nomad/structs/csi_volumes_test.go (Browse further)