Diagnostic code

ZC-E0515

a swarm fetch outlived its kill, so the image was not pulled

Kind

error

Blame

environment

Exit code

2

What it means

A replica's image was being fetched from peers into this node's Zigbox store, the fetch passed its deadline (or ended badly), and the supervisor killed the fetch's whole process group with SIGKILL. A process of that group was still alive afterwards. It may still be writing the store, and Zigbox's store takes one writer at a time, so the supervisor did not start the registry pull: the replica was not started.

A process that outlives SIGKILL is in uninterruptible sleep, waiting on the kernel: usually a disk or a network mount that stopped answering. Find it (ps -eo pid,stat,wchan:32,args, state D) and the device it waits on. The replica is started again on a later pass once it is gone.

Offline

The same text, without a network:

zcube explain E0515