Planner fails when one machine is not available

[ndias]

I don't know since, but I think in previous versions of amanda this wouldn't happen.

When I have a machine that for unknown reason can't be reachable the planner fails for all machines and the backups are aborted for all machines.

FAILURE DUMP SUMMARY:
planner: ERROR Request to x50 failed: EOF on read from x50
x1 disk1 RESULTS MISSING
x2 disk2 RESULTS MISSING
...
x100 disk100 RESULTS MISSING

NOTES:
planner: tapecycle (100) <= runspercycle (100)
driver: WARNING: got empty schedule from planner

The version of amanda is 3.5.4

[LouHafer]

Just filed a bug report with Fedora Bugzilla for what might be your issue. For me it's been a problem since I upgraded to Fed38. If a host was down, I'd get the EOF on read message and planner would abort. Same for amcheck. Solution for me is a patch to common-src/security-utils.c:sec_tcp_conn_read_callback. On bugzilla.redhat.com, check bug 2342328.