Skip to content

Read pipeline logs

Use saif pipeline logs <build> to read logs without rerunning a build. Start with the failing step, then inspect earlier jobs in the same run when a later lookup says a resource or artifact does not exist.

Reading Build Logs from the CLI

Mode Input Context
Remote Full build URL The URL supplies the host, collection, project, and build ID
Contextual Build ID Repository selector, explicit coordinates, current repository, or interactive selection

Remote mode (by URL)

Paste the build URL from the browser:

saif pipeline logs "https://dev.azure.com/SAIFCorporation/Customer/_build/results?buildId=162774"

Do not combine a URL with --host, --collection, --project, or --repo; remote mode already has its coordinates and rejects those context flags.

A URL copied from a stage, job, or step can include si, j, or t. Those identifiers scope the output to that node. Use --all-steps to remove the URL-derived scope and inspect the whole build:

saif pipeline logs "https://dev.azure.com/SAIFCorporation/Customer/_build/results?buildId=162774" --all-steps

For example, if Terraform apply fails, inspect the preceding Check Terraform Cloud Workspace Existence step before assuming the workspace is missing. The check logs the resolved workspace name and ID. Compare them with the apply step's target and error; a successful existence check rules out that particular missing-workspace diagnosis, not every later Terraform failure.

Contextual mode (by id)

From a repository clone, the CLI can infer context from its git remote. Outside a clone, select a repository or provide coordinates:

saif pipeline logs 162774
saif pipeline logs 162774 --repo my-service
saif pipeline logs 162774 --host dev.azure.com --collection SAIFCorporation

saif pipeline monitor <build-id> resolves a bare build ID through the same context resolution, so it also infers the repository from the current clone.

Filtering and output

# Only failed steps
saif pipeline logs 162774 --status failed

# Scope to a stage and step
saif pipeline logs 162774 --stage Build --step Compile

# Last 50 lines across matching steps
saif pipeline logs 162774 --tail 50

# Structured output
saif pipeline logs 162774 --format json

Filters are case-insensitive and combinable. --stage, --job, and --step accept a name or timeline-record GUID; --stage also accepts the stage's YAML identifier. Explicit filters override the corresponding URL-derived scope. --all-steps ignores URL scope, not filters you explicitly supplied.

For a running build, --follow streams new lines until completion. It cannot combine with --tail or machine-readable output (json, jsonlines, or plain). A log command can return a failure exit code because the selected build steps failed, even when it successfully retrieved their logs.

Compare earlier and later jobs

  1. Identify the first failing operation: initialization, checkout, restore/build, test, or deployment.
  2. If it consumes a resource from an earlier job, read that job's result and recorded identifier.
  3. Compare environment, workspace, resource name, and lookup filter. A successful create followed by an empty lookup suggests a lookup or timing problem, not proof that creation never ran.
  4. Preserve the build URL and relevant error before changing configuration. Redact tokens and sensitive log values before sharing.

See Troubleshooting for symptom-specific articles, and Pipelines for the pipeline and variable contracts.

Command behavior comes from PipelineLogsCommand. Use saif pipeline logs --help for the installed CLI's complete option surface.