We have a debian stretch backup server running bacula 7.4.4 (fd, sd and director all same version). I tried to connect some new servers running debian buster with bacula fd 9.4.2 with no success. When i try to run a job, the following error appears in the logs:
...
[backuphost]-sd JobId 6802: Fatal error: bsock.c:569 Packet size=1073741933 too big from "client:[ip]:9103. Terminating connection.
[backuphost]-dir JobId 6802: Error: getmsg.c:185 Malformed message: Jmsg JobId=6802 type=4 level=1577934200 02-Jan 04:03 [backuphost]-dir JobId 6802: Start Backup JobId 6802, Job=[client]_backup.2020-01-02_04.00.00_11
[backuphost]-sd JobId 6802: Elapsed time=00:00:01, Transfer rate=162 Bytes/second
[backuphost]-dir JobId 6802: Error: getmsg.c:185 Malformed message: Jmsg JobId=6802 type=3 level=1577934200 [client] JobId 6802: Fatal error: backup.c:929 Network send error to SD. Data=51 1 0 ERR=Connection reset by peer
[backuphost]-dir JobId 6802: Error: getmsg.c:185 Malformed message: Jmsg JobId=6802 type=4 level=1577934210 [client] JobId 6802: Error: bsock.c:271 Socket has errors=1 on call to Storage daemon:[hostip]:9103
[backuphost]-dir JobId 6802: Fatal error: bsock.c:569 Packet size=1073741944 too big from "Client: [client]:[clientip]:9102. Terminating connection.
[backuphost]-dir JobId 6802: Fatal error: No Job status returned from FD.
It is not a hardware error or network/firewall misconfiguration, because all the other servers running 7.4.4 are working, nmap shows that host->client 9102 is open, client->host 9103 is open on all related clients, and the error always appears on the 3 clients running fd 9.4.2.
After some search, i tried some configuration options like - Changing "Maximum Network Buffer Size" in fd and sd - Changing "Heartbeat Interval" in fd and sd Same error each time.
As a temporary workaround, i downgraded the fd version to 7.4.4 on a client and the backup works with the configuration file unchanged.
If this is a simple configuration difference between the versions, i'd really like to know how to fix it, if its a compatibility error, than the dev team should know about it. The release notes only mentions lower fd version compatibility, not higher.
Long term solution would be a host upgrade, but i'm reluctant to do it at this point.
