* created roadmap and yaml claude agent
* Update roadmap.md
* feat: Clean advanced analysis implementation - core agents, engine, artifacts
* Remove unrelated files - keep only advanced analysis implementation
* fix: Fix goroutine leak in hosted agent rate limiter
- Added stop channel and stopped flag to RateLimiter struct
- Modified replenishTokens to listen for stop signal and exit cleanly
- Added Stop() method to gracefully shutdown rate limiter
- Added Stop() method to HostedAgent to cleanup rate limiter on shutdown
Fixes cursor bot issue: Rate Limiter Goroutine Leak
* fix: Fix analyzer config and model validation bugs
Bug 1: Analyzer Config Missing File Path
- Added filePath to DeploymentStatus analyzer config in convertAnalyzerToSpec
- Sets namespace-specific path (cluster-resources/deployments/{namespace}.json)
- Falls back to generic path (cluster-resources/deployments.json) if no namespace
- Fixes LocalAgent.analyzeDeploymentStatus backward compatibility
Bug 2: HealthCheck Fails Model Validation
- Changed Ollama model validation from prefix match to exact match
- Prevents false positives where llama2:13b would match request for llama2:7b
- Ensures agent only reports healthy when exact model is available
Both fixes address cursor bot reported issues and maintain backward compatibility.
* fixing lint errors
* fixing lint errors
* adding CLI flags
* fix: resolve linting errors for CI
- Remove unnecessary nil check in host_kernel_configs.go (len() for nil slices is zero)
- Remove unnecessary fmt.Sprintf() calls in ceph.go for static strings
- Apply go fmt formatting fixes
Fixes failing lint CI check
* fix: resolve CI failures in build-test workflow and Ollama tests
1. Fix GitHub Actions workflow logic error:
- Replace problematic contains() expression with explicit job result checks
- Properly handle failure and cancelled states for each job
- Prevents false positive failures in success summary job
2. Fix Ollama agent parseLLMResponse panics:
- Add proper error handling for malformed JSON in LLM responses
- Return error when JSON is found but invalid (instead of silent fallback)
- Add error when no meaningful content can be parsed from response
- Prevents nil pointer dereference in test assertions
Fixes failing build-test/success and build-test/test CI checks
* fix: resolve all CI failures and cursor bot issues
1. Fix disable-ollama flag logic bug:
- Remove disable-ollama from advanced analysis trigger condition
- Prevents unintended advanced analysis mode when no agents registered
- Allows proper fallback to legacy analysis
2. Fix diff test consistency:
- Update test expectations to match function behavior (lines with newlines)
- Ensures consistency between streaming and non-streaming diff paths
3. Fix Ollama agent error handling:
- Add proper error return for malformed JSON in LLM responses
- Add meaningful content validation for markdown parsing
- Prevents nil pointer panics in test assertions
4. Fix analysis engine mock agent:
- Mock agent now processes and returns results for all provided analyzers
- Fixes test expectation mismatch (expected 8 results, got 1)
Resolves all failing CI checks: lint, test, and success workflow logic
---------
Co-authored-by: Noah Campbell <noah.edward.campbell@gmail.com>
* Auto-collectors: foundational discovery, image metadata, CLI integration; reset PRD markers
* Address PR review feedback
- Implement missing namespace exclude patterns functionality
- Fix image facts collector to use empty Data field instead of static string
- Correct APIVersion to use troubleshoot.sh/v1beta2 consistently
* Fix bug bot issues: API parsing, EOF error, and API group corrections
- Fix RBAC API parsing errors in rbac_checker.go (getAPIGroup/getAPIVersion functions)
- Fix FakeReader EOF error to use standard io.EOF instead of custom error
- Fix incorrect API group from troubleshoot.sh to troubleshoot.replicated.com in run.go
These changes address the issues identified by the bug bot and ensure proper
interface compliance and consistent API group usage.
* Fix multiple bug bot issues
- Fix RBAC API parsing errors in rbac_checker.go (getAPIGroup/getAPIVersion functions)
- Fix FakeReader EOF error to use standard io.EOF instead of custom error
- Fix incorrect API group from troubleshoot.sh to troubleshoot.replicated.com in run.go
- Fix image facts collector Data field to contain structured JSON instead of static strings
These changes address all issues identified by the bug bot and ensure proper
interface compliance, consistent API usage, and meaningful data fields.
* Update auto_discovery.go
* Fix TODO comments in Auto-collector section
Fixed 3 of 4 TODOs as requested in PR review:
1. pkg/collect/images/registry_client.go (line 46):
- Implement custom CA certificate loading
- Add x509 import and certificate parsing logic
- Enables image collection from private registries with custom CAs
2. cmd/troubleshoot/cli/diff.go (line 209):
- Implement bundle file count functionality
- Add tar/gzip imports and getFileCountFromBundle() function
- Properly counts files in support bundle archives (.gz/.tgz)
3. cmd/troubleshoot/cli/run.go (line 338):
- Replace TODO with clarifying comment about RemoteCollectors usage
- Confirmed RemoteCollectors are still actively used in preflights
The 4th TODO (diff.go line 196) is left as-is since it's explicitly marked
as Phase 4 future work (Support Bundle Differencing implementation).
Addresses PR review feedback about unimplemented TODO comments.
---------
Co-authored-by: Benjamin Yang <benjaminyang@Benjamins-MacBook-Pro.local>
* Allowing more comperators for the http analyzer
* test
* Update pkg/analyze/host_http.go
Co-authored-by: Andrew Lavery <laverya@umich.edu>
---------
Co-authored-by: Andrew Lavery <laverya@umich.edu>
The OS version analyzer did not allow checking for things like "redhat 8.x" - this equates to >= 8 && < 9 in the new code.
Also, we previously only collected the OS name (like redhat, centos, or ubuntu) not the OS family (which would be rhel, rhel, and debian for the previous OSes) - this greatly reduces the number of cases required in an analyzer.
* feat: implement collector and analyser for network namespace connectivity
checks if two network namespaces can talk to each other on udp and tcp.
its usage is as follows:
```yaml
apiVersion: troubleshoot.sh/v1beta2
kind: SupportBundle
metadata:
name: test
spec:
hostCollectors:
- networkNamespaceConnectivity:
collectorName: check-network-connectivity
fromCIDR: 10.0.0.0/24
toCIDR: 10.0.1.0/24
hostAnalyzers:
- networkNamespaceConnectivity:
collectorName: check-network-connectivity
outcomes:
- pass:
message: "Communication between 10.0.0.0/24 and 10.0.1.0/24 is working"
- fail:
message: "Communication between 10.0.0.0/24 and 10.0.1.0/24 isn't working"
```
if this fails then you may need to enable `forwarding` with:
```bash
sysctl -w net.ipv4.ip_forward=1
```
if it still fails then you may need to configure firewalld to allow the
traffic or simply disable it for sake of testing.
* chore: rebuild schemas
* chore: remove unused property
* chore: disable namespaces for other platforms
* chore: make sure we timeout temporary servers
* feat: analyzer now supports multi-node collection
* feat: check both udp and tcp even on failure
check both protocols even if one fails. this pr commit also introduces a
timeout that can be set by the user.
* feat: add templating to the failure outcome
allow users to dump the errors found during the analysis.
* chore: addressing pr comments
* feat: delete interface pair before namespace
even though the interface pair is deleted everyttime we delete the
namespace on my tests we better delete it before we delete the
namespace.
this comes out of a review comment where some people seem to still be
able to see the interface pair even after the namespace is deleted.
i.e. better safe than sorry.
* chore: fix typo on comment
allow users to check if specific cpu flags are supported by the host.
```yaml
apiVersion: troubleshoot.sh/v1beta2
kind: HostPreflight
metadata:
name: ec-cluster-preflight
spec:
collectors:
- cpu: {}
analyzers:
- cpu:
checkName: CPU
outcomes:
- pass:
when: hasFlags cmov,cx8,fpu,fxsr,mmx
message: CPU supports all required flags
- fail:
message: CPU not supported
```
allows troubleshoot to collect and analyze CPU micro architecture. this
is an usage example:
```yaml
apiVersion: troubleshoot.sh/v1beta2
kind: HostPreflight
metadata:
name: ec-cluster-preflight
spec:
collectors:
- cpu: {}
analyzers:
- cpu:
checkName: CPU
outcomes:
- pass:
when: 'supports x86-64-v2'
message: CPU supports x86-64-v2
- fail:
message: CPU does not support x86-64-v2
```
Change to stop re-analysing preflight results when uploadResultsTo is present leading to duplicate results
Signed-off-by: Evans Mungai <evans@replicated.com>
* feat: Install goldpinger if one does not exist when running goldpinger collector
- Deploy golpinger daemonset if one is not detected in the cluster
- Clean up all deployed resources
- Add delay to allow users to wait for goldpinger to perform checks
Signed-off-by: Evans Mungai <evans@replicated.com>
* Add missing test data file
Signed-off-by: Evans Mungai <evans@replicated.com>
* Better naming of create resource functions
Signed-off-by: Evans Mungai <evans@replicated.com>
---------
Signed-off-by: Evans Mungai <evans@replicated.com>
* new struct and update schemas
* implement Collect function
* add kernel config to collector struct
* generate kernel config analyzer schema
* implement kernel config analyzer
* fail on no match in pass outcome
* run make check-schemas
* fix failed unit test
* update from code review
* add selectedConfigs field
* run make check-schemas
* feat: node metrics analyser
The analyser only checks PVC usage at the moment. More analysers
can be added on a need to have basis
* Add tests
* Fix flaky test by waiting for goldpinger pods to start
* Fix how outcomes get checked
* Fix catch all outcome condition
* Fix test
* feat: node metrics analyser
The analyser only checks PVC usage at the moment. More analysers
can be added on a need to have basis
* Add tests
* Fix flaky test by waiting for goldpinger pods to start
* Fix how outcomes get checked
* Fix catch all outcome condition
* Fix test
* Regenerate schemas
* Fix failing test
---------
Co-authored-by: Dexter Yan <yanshaocong@gmail.com>
AshandGitHub
Diamon WigginsandGitHub
Ethan MosbaughandGitHub
Andrew LaveryandGitHub
Johannes TuchschererandGitHub
Greg SchofieldandGitHub
Salah Al SalehandGitHub
Gerard NguyenandGitHub
Miguel Varela RamosandGitHub
João AntunesandGitHub
Ricardo MaraschiniandGitHub
Shubhag SaxenaandGitHub