While working through T385896 , I noticed that wdqs-categories update lag alerts are being routed to the mainline SRE team. There are arguments for and against routing this tickets to mainline SRE,
FOR
- The internal wdqs-categories service hosts the subcategory graphs for production wikis. It can be mitigated pretty easily by running the wdqs-categories-reload cookbook.
AGAINST
- Mainline SREs haven't specifically been trained for this, so they might not feel comfortable responding.
- Lag on a single host is not disruptive enough to wiki user experience (subjective).
Creating this ticket to:
- Discuss alert routing with stakeholders (mainline SRE, Search Platform, Data Platform SRE).
- Route tickets according to agreed-upon plan (my personal preference is to route these to Data Platform SRE only)
PS This discussion implies that wdqs-categories should have an SLO, we should probably discuss this as well.