Seems we have currently a lot of backlog on try and partly on inbound for tegras doing tests. Nick asked me to file a bug to investigate
raising servertiy, according to the sheriff notes: backlog got up around 11 hours during the day, but extensive multi-tree closures are driving it back down until the next time we open trees, or tomorrow morning)
The backlog across all test types was exceptional last week at many times. I account for that due to an increase in the amount of AWS disconnects we saw, (many per day, rather than just one or two per week) which translates into RETRY for all tests it hits. We also have a limited tegra pool, and tegras are typically on our long pole of test types, couple that with there having been a handful of tegras out of the pool awaiting reimage, and we get into this state. Nothing more we can investigate/do here though.