Clojurians Log v2

we have a cluster running 0.9.x and one running 0.14.x that we are migrating our existing jobs to. On the 0.9.x cluster, we were tracking down some rare, but annoying issues that kill jobs. upon closer inspection, it turns out we had segments from tasks in one job winding up in tasks in different jobs, not specific to any one job. The jobs themselves don’t share any inputs or outputs and the only way I can conceive that these segments wound up in a different job entirely is via Aeron, either publishing to the wrong place or consuming the wrong message. has anyone experienced anything similar? It does seem to happen very rarely, and we have yet to see it in our 0.14.x cluster.