Monday was going to be a full work stoppage.
Two problems had reached the point where everything else needed to wait. We had recurring overnight outages caused by load in a critical part of the system. A separate dependency was blocking another important path. Both were affecting the product. Neither was going to improve while we kept treating it as one more item beside the normal work.
The team had already been responding. People traced the latest outage, paused the work that appeared to be triggering it, and started looking at options. That bought us some room, but it was not a resolution. This was happening repeatedly, and repeated outages turn every morning into an investigation of the night before.
My Friday note was blunt: stop all other work Monday until these two items are addressed.
I knew what that meant. Planned product work would stop. Architecture work would stop. People who had their own priorities would be pulled together around problems they might not have created. We would lose a day, maybe more, against the plans already on the wall.
The alternative was worse. A roadmap does not matter much if the product cannot stay available. New features do not help if a core path is blocked. Continuing normal work would make the schedule look healthier while the actual system stayed sick.
I asked the technical owners to drive both the communication and the path to resolution. The rest of us needed to team up, solve the blockers, and then get back to the work we had interrupted.
There was no point assigning blame in the outage thread. One issue came from load inside our own process. Another involved something outside the immediate application. The useful question was what had to change so neither could keep stopping us.
Friday ended with the priorities reset. Monday would not begin with the regular backlog. It would begin with two problems, everybody needed to fix them, and nothing else got to be more important.