* add leases path to sudo required set
* update TestSystemBackend_RootPaths with new special privilege paths
* note that list-leases requires sudo
* minor typo fixes
* diagnose: Add seal transit tls check
* Fixing the path to the config file and the path to the cert files
* Addressing comment
* Addressing seal transit tls check comments
* tls verification bugfix
* tls verification bugfix
* allow diagnose fail to report status when there are also warnings
* allow diagnose fail to report status when there are also warnings
* Update vault/diagnose/helpers_test.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* comments
Co-authored-by: swayne275 <swayne275@gmail.com>
* Checking Validity of all Certs in the chain
* Addressing Comments for TLS cert validation
* Fixing tls_verification tests
* Fixing minor issue in tls_verification tests
* Addressing Comments, Rebasing with main
* Adding comment on top of a test
* Actually call config.Validate in diagnose
* Wire configuration checks into diagnose and fix resulting bugs.
* go mod vendor
* Merge to vendorless version
* Remove sentinel section to allow diagnose_ok to pass
* Fix unit tests
* raft file and quorum checks
* raft checks
* backup
* raft file checks test
* address comments and add more raft and file and process checks
* syntax issues
* modularize functions to compile differently on different os
* compile raft checks everywhere
* more build tag issues
* raft-diagnose
* correct file permission checks
* upgrade tests and add a getConfigOffline test that currently does not work
* comment
* update file checks method signature on windows
* Update physical/raft/raft_test.go
Co-authored-by: Brian Kassouf <briankassouf@users.noreply.github.com>
* raft tests
* add todo comment for windows root ownership
* voter count message
* raft checks test fixes
Co-authored-by: Brian Kassouf <briankassouf@users.noreply.github.com>
* initial refactoring of unseal step in run
* remove waitgroup
* remove waitgroup
* backup work
* backup
* backup
* completely modularize run and move into diagnose
* add diagnose errors for incorrect number of unseal keys
* comment tests back in
* backup
* first subspan
* finished subspanning but running into error with timeouts
* remove runtime checks
* merge main branch
* meeting updates
* remove telemetry block
* roy comment
* subspans for seal finalization and wrapping diagnose latency checks
* backup while I fix something else
* fix storage latency test errors
* runtime checks
* diagnose with timeout on seal
* Acquire a per-lock lease to make renew and revoke atomic wrt each other.
This means we don't have to hold pendingLock during I/O.
* Attempted fix for deadlock in token revocation.
* Comment fix.
* Fix error checking in loadEntry.
* Add benchmark
* Add a few additional locking locations
* Improve benchmark slightly
* Update vault/expiration.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* Update vault/expiration.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* Add a lease lock into tidy
Co-authored-by: Scott Miller <smiller@hashicorp.com>
Co-authored-by: Brian Kassouf <bkassouf@hashicorp.com>
Co-authored-by: Brian Kassouf <briankassouf@users.noreply.github.com>
Co-authored-by: swayne275 <swayne275@gmail.com>
* wip
* wip
* Finish implementing advice handling and word wrapping
* Properly word wrap messages and warnings
* Remove debugging
* Remove debugging
* Remove unnecessary test
* unit test bug
* go vendor
* Segment out disk checks to disable on openbsd/arm
Also add a spot skipped helper.
* Expected results may be fewer than actual because of variable length tests like disk usage
* Move to os_common and build on windows
* build out lease count (not fully working), start lease list
* build out irrevocable lease list
* bookkeeping
* test irrevocable lease counts for API/CLI
* fix listIrrevocableLeases, test listIrrevocableLeases, cleanup
* test expiration API limit
* namespace tweaks, test force flag on lease list
* integration test leases/count API, plenty of fixes and improvements
* test lease list API, fixes and improvements
* test force flag for irrevocable lease list API
* i guess this wasn't saved on the last refactor...
* fixes and improvements found during my review
* better test error msg
* Update vault/logical_system_paths.go
Co-authored-by: Brian Kassouf <briankassouf@users.noreply.github.com>
* Update vault/logical_system_paths.go
Co-authored-by: Brian Kassouf <briankassouf@users.noreply.github.com>
* return warning with data if more than default leases to list without force flag
* make api doc more generalized
* list leases in general, not by mount point
* change force flag to include_large_results
* sort leases by LeaseID for consistent API response
* switch from bool flag for API limit to string value
* sort first by leaseID, then stable sort by expiration
* move some utils to be in oss and ent
* improve sort efficiency for API response
Co-authored-by: Brian Kassouf <briankassouf@users.noreply.github.com>
* Add ulimit check, and tidy unit test cases to avoid needing to have all results and perfect ordering
* Make order independent check recursive
* Fix unit tests
* Try a 5s request timeout
Checking if a plugin is a builtin by comparing it's type to those in the builtin
registry allows for a custom plugin loaded with the same name to be considered a
builtin during error handling of the mounting process. This can cause the vault
installation to brick itself because it cannot be unsealed to register a new
sha256 or file path for a previously loaded custom plugin. Improve this logic
by checking the plugin catalog rather than the builtin registry.
Fixes#11687
* Disk usage checks
* Move disk free earlier
* Move logic to helpers
* Bring over test logic from the ulimit PR
* imports
* Report error
* Get unit tests working
* initial refactoring of unseal step in run
* remove waitgroup
* remove waitgroup
* backup work
* backup
* backup
* completely modularize run and move into diagnose
* add diagnose errors for incorrect number of unseal keys
* comment tests back in
* backup
* first subspan
* finished subspanning but running into error with timeouts
* remove runtime checks
* meeting updates
* remove telemetry block
* roy comment
* subspans for seal finalization and wrapping diagnose latency checks
* fix storage latency test errors
* review comments
* use random uuid for latency checks instead of static id
* Create helpers which integrate with OpenTelemetry for diagnose collection
* Go mod vendor
* Comments
* Update vault/diagnose/helpers.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* Add unit test/example
* tweak output
* More comments
* add spot check concept
* Get unit tests working on Result structs
* wip
* Fix unit test
* Get unit tests working, and make diagnose sessions local rather than global
* Comments
* Last comments
* No need for init
* :|
* Fix helpers_test
* wip
* wip
* wip
* Revendor otel
* Fix merge related problems
* imports
* Fix unit tests
Co-authored-by: swayne275 <swayne275@gmail.com>
* Add infrastructure for skipping tests
* Add infrastructure for skipping tests
* Set it
* Update vault/diagnose/helpers.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* Implement type alias for test functions
Co-authored-by: swayne275 <swayne275@gmail.com>
* shrink generic error message
* move zombie loading to updatePendingInternal from loadEntryInternal
* probably the right metric/lease behavior for irrevocable leases...
* comment improvements
* test total lease count with valid and irrevocable leases
* The main driver for this change was to make the read from a.newFragmentCh timeout quickly rather than waiting for the test timeout (much longer). While testing the change I observed a panic during shutdown, but it was swallowed and moreover there was no stack trace so it wasn't obvious. I'm hoping we can get rid of the recover, so I fixed the issue in the activitylog tests that needed it.
* Add an Int64 type
* Use the new Int64 type so that even 32 bit builds can specify max_operations above 2^31
* Missed a spot
* go mod vendor
* fix cast
* changelog
* Update unit test to ensure this works on both 32 and 64-bit archs
* build out zombie lease system
* add typo for CI
* undo test CI commit
* time equality test isn't working on CI, so let's see what this does...
* add unrecoverable proto error, make proto, go mod vendor
* zombify leases if unrecoverable error, tests
* test fix: somehow pointer in pointer rx is null after pointer rx called
* tweaks based on roy feedback
* improve zombie errors
* update which errors are unrecoverable
* combine zombie logic
* keep subset of zombie lease in memory
* Create helpers which integrate with OpenTelemetry for diagnose collection
* Go mod vendor
* consul tls checks
* draft for storage end to end check
* Comments
* Update vault/diagnose/helpers.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* Add unit test/example
* tweak output
* More comments
* add spot check concept
* Get unit tests working on Result structs
* Fix unit test
* Get unit tests working, and make diagnose sessions local rather than global
* Comments
* Last comments
* No need for init
* :|
* Fix helpers_test
* cleaned up chan logic. Tests next.
* fix tests
* remove a comment
* tests
* remove a comment
* run direct access checks in diagnose command
* review comments
Co-authored-by: Scott G. Miller <smiller@hashicorp.com>
Co-authored-by: swayne275 <swayne275@gmail.com>
* Create helpers which integrate with OpenTelemetry for diagnose collection
* Go mod vendor
* consul tls checks
* draft for storage end to end check
* Comments
* Update vault/diagnose/helpers.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* Add unit test/example
* tweak output
* More comments
* add spot check concept
* Get unit tests working on Result structs
* Fix unit test
* Get unit tests working, and make diagnose sessions local rather than global
* Comments
* Last comments
* No need for init
* :|
* Fix helpers_test
* cleaned up chan logic. Tests next.
* fix tests
* remove a comment
* tests
* remove a comment
* cosmetic changes
Co-authored-by: Scott G. Miller <smiller@hashicorp.com>
Co-authored-by: swayne275 <swayne275@gmail.com>
* Create helpers which integrate with OpenTelemetry for diagnose collection
* Go mod vendor
* Comments
* Update vault/diagnose/helpers.go
Co-authored-by: swayne275 <swayne275@gmail.com>
* Add unit test/example
* tweak output
* More comments
* add spot check concept
* Get unit tests working on Result structs
* Fix unit test
* Get unit tests working, and make diagnose sessions local rather than global
* Comments
* Last comments
* No need for init
* :|
* Fix helpers_test
Co-authored-by: swayne275 <swayne275@gmail.com>
* build out zombie lease system
* add typo for CI
* undo test CI commit
* time equality test isn't working on CI, so let's see what this does...
* protect against nil pointer receiver calls
* new font and add as font-family to be used in masked-input
* clean up logic
* refactor for displayOnly
* start cert masking
* work on certificates
* upload cert work
* fix global styling
* fix styling for class no longer used
* make mask by default and remove option
* glimmerize start and certificate on LDAP a file field
* glimmerize actions
* first part of glimmerizing text-file still need to do some clean up
* not doing awesome over here
* getting ready to un-glimmer
* unglimmerize
* remove placeholder based on conversations with design
* clean up text-file
* cleanup
* fix class bindings
* handle class binding
* set up for test
* fix elementId
* track down index
* update masked-input test
* add more to the masked-input test
* test-file test
* fix broken test
* clear old style
* clean up
* remove pgp key masked font, this really needs to be refactored to text-file component
* changelog
* cover other certificate view
* add allowCopy
* address some pr styling comments
* improve test coverage
* fix some issues
* add attr.options.masked
* Add support for unauthenticated pprof access on a per-listener basis, as we do for metrics.
* Add missing pprof sub-targets like 'allocs' and 'block'. Capture the goroutine subtarget a second time in text form. This is mostly a convenience, but also I think the pprof format might be a bit lossy?
The rpcClientConnContext is still used to terminate gRPC internal/dialer-related goroutines, but the actual RPC is now timed out when the request times out, e.g. due to the default max request duration. This mirrors what we do with the parallel forwarding code in ENT.
* tls tests and root verification
* make the certificate verification check correct for non root CA case
* add expiry test
* addressed comments but struggling with the bug in parsing Cas and inters from single file:
* final checks on tls and listener
* cleanup
* sanity checks for tls config in diagnose
* backup
* backup
* backup
* added necessary tests
* remove comment
* remove parallels causing test flakiness
* comments
* small fix
* separate out config hcl test case into new hcl file
* newline
* addressed comments
* addressed comments
* addressed comments
* addressed comments
* addressed comments
* reload funcs should be allowed to be nil
* Boost max_operations to the greater of that specified or absoluteMinOperations
* Forward rotation config requests to the primary
* Reject rotation configs outside the min/max range
* Minor wording fix
swayne275