Yocto Build Failure Swat Team

Overview

The role of the SWAT team is to monitor the autobuilder and investigate all failures to ensure that they are logged and brought to the attention of the appropriate owner.

All builds that are run on the public autobuilder are important for the Yocto Project, whether they be a post-merge validation run (for master or a release branch) or a pre-merge test build (for master-next, ross/mut, and others).

Every build should be monitored by the SWAT team unless the BuildLog entry for that build indicates otherwise. SWAT is opt-out by whomever triggers a build on the autobuilder, not opt-in.

SWAT isn't responsible for resolving any issues encountered on the autobuilder. Their focus is on performing minimal analysis of a failure in order to ensure that it is logged and brought to the attention of a appropriate owner, a process we'll refer to as pre-triage.

Rotation Process

The active member rotation takes place weekly at the end of Friday. Usually, this will take a simple round robin order through the members list. In case the next person cannot take the role due to tight schedule, vacation or some other reasons, the role will be passed to the next person.

Roles

Active member: the currently active member of the SWAT team is expected to monitor the autobuilder and pre-triage failures in a timely fashion. Team members are active for one week at a time.

SWAT Chair: the SWAT chair provides backup cover for the active member and is a first point of contact for SWAT. Ross Burton is the current SWAT Chair.

SWAT Facilitator: the SWAT facilitator is responsible for managing the rotation process. Stephen K. Jolley is the current SWAT Facilitator.

Process

The BuildLog wiki page is automatically updated by the autobuilder when a new build is triggered. This often includes guidance for SWAT (the "Reason" field when triggering the build) such as to ignore it (if the build is expected to break and the owner is monitoring themselves) or report failures straight to a specific person (if the build only contains changes from a single person, such as a GCC upgrade). Unless when told otherwise, the usual process is as follows:

For builds against master or a release branch, any issues observed should be filed in Bugzilla.

For builds against other branches (master-next, ross/mut, -next branches for stable releases, etc.), attempt to identify what patch in the branch is likely responsible for the failure. For example, if wget fails with libgnutls errors and there is a GnuTLS upgrade in the branch, that is the likely candidate. If a patch can be identified that isn't also in master or release branch, reply on the mailing list with the failure details. If it isn't obvious which patch is responsible for the failure then file a bug and ensure the branch owner is either the assignee or on the CC list.

If in doubt, file a bug. All observed errors must be actioned unless a patch has already been sent for the issue, in which case please make note of this in the BuildLog.

If the issue is in the infrastructure or autobuilder itself then file a bug against , infrastructure bugs should be assigned to Michael Halstead and autobuilder logic bugs to User:Rpurdie.

The results of pre-triage for an issue should be added to the corresponding entry in the BuildLog, including a link to the resolution (patch name, bug link, etc) and a brief summary of the issue.

Every build failure should be addressed in the BuildLog. If it is a known issue, an entry with a single line containing "Known Issue" is sufficient (a link to further detail is, of course, much better). This assures others that the failure has been looked at and is being worked on.

Filing bugs

When filing the bug, please:

• cut and paste the relevant error in the bug comment, and include the log file as an attachment
• include the log from the CreateAutoConf step as an attachment (this ensures the assignee and triage team can quickly asses this issue)
• include a pointer to the ErrorLog page associated with the failure (as ErrorLog)

Note: Autobuilder logs are non-persistent, feel free to include a link to the log in a bug report but be sure to also attach a copy of the log and include relevant sections copy/pasted into the bug.

Note: Sometimes, failures occur on autobuilders on private company networks. Do not post links into the bugzilla for these failures as nobody else can access them.

Process summary

• Monitor builds via one (or more) of:
  • the autobuilder: main page, console view.
  • the BuildLog wiki page
  • the yocto-builds mailing list
• Pre-triage each failure:
  • File a bugzilla ticket OR respond to a patch OR note known issues
  • Update the BuildLog with the result of pre-triage, linking to issues/mail archives when possible

Questions / Contact

If you have queries about the SWAT process you may reach out to the SWAT Facilitator Stephen K. Jolley and the SWAT Chair Tracy Graydon.

Members

• Paul Eggleton (NZ)
• Ross Burton (UK)
• Amanda Brindle (US)
• Chen Qi (CN)
• Armin Kuster (California)