The goal of the support playbook is to spread knowledge about how to support MongoDB running WiredTiger.
We're going to try to work towards a common format where each playbook entry is a separate wiki page describing a class of problem with:
-
symptoms
-
background
-
diagnosis and remediation
Types of issues to cover:
-
Performance issues (tools e.g:
timeseries) -
Cache usage problems
-
Machine configuration problems, identifying them and outlining solutions
-
Crashes that involve a stack trace (search JIRA, can we have a protocol to make tickets tracking crashes reliably searchable)
-
Issues that mean recovery can’t startup a
mongod(e.g., checksum mismatches)
Process:
-
Use JIRA to find common / important cases and use them to drive the doc
-
Add a “playbook” label to JIRA tickets that have relevant information
-
Add a line to the tracking page
-
Add a developer-only comment with notes about why / what should go in the playbook
-
Ultimate output will be on the wiki, arranged so that relevant playbook pages turn up in searches