This happens quite a bit - which is rather visible in Staging tests.
It happens on x86_64, but seems much more predominant on i586.
e.g.
osc jobhist Virtualization/qemu/openSUSE_Tumbleweed/i586
2022-09-10 00:14:07 qemu meta
change unchanged 3h 32m 16s cloud126:2
2022-09-10 03:04:47 qemu rebuild
counter succeeded 2h 50m 34s cloud108:1
2022-09-11 07:55:46 qemu meta
change failed 10h 28m 33s lamb24:8
2022-09-12 00:00:30 qemu meta
change succeeded 1h 4m 30s sheep88:12
2022-09-13 06:46:07 qemu meta
change failed 11h 12m 23s build83:2
2022-09-14 18:55:48 qemu meta
change unchanged 2h 11m 58s lamb59:10
2022-09-14 20:09:29 qemu rebuild
counter succeeded 1h 13m 36s lamb68:3
2022-09-15 17:17:01 qemu source
change succeeded 2h 21m 18s build77:2
2022-09-16 11:00:29 qemu rebuild
counter failed 9h 43m 10s lamb22:6
2022-09-16 23:39:26 qemu meta
change failed 9h 29m 32s lamb66:2
2022-09-17 09:11:49 qemu meta
change failed 9h 32m 15s lamb67:1
2022-09-17 23:01:05 qemu meta
change succeeded 41m 29s stones:13
2022-09-19 07:21:45 qemu meta
change failed 9h 32m 50s sheep88:8
2022-09-19 14:31:01 qemu meta
change unchanged 1h 53m 36s lamb06:2
2022-09-20 22:41:41 qemu meta
change failed 9h 30m 48s lamb64:4
2022-09-21 00:11:33 qemu meta
change unchanged 1h 29m 47s lamb64:4
2022-09-21 01:28:10 qemu rebuild
counter succeeded 1h 16m 31s lamb63:2
As build already takes long, it would be much appreciated if they'd not need to
be retried more often than really needed