I just upgraded Amanda on two systems here, server and client, from
version 2.4.x to 2.5.2_p1 to avoid a number of problems. Unfortunately,
I'm no longer able to connect with the amanda client from my server,
although I was previously able to do so. Both the client and server
were upgraded to the new version of amanda and amanda.conf and
amanda-client.conf were inspected and upgraded on their respective
machines.
The client and server are at opposite ends of a VPN tunnel, and all IP
addresses involved are those of the VPN interfaces.
... From the server:
Amanda Backup Client Hosts Check
--------------------------------
WARNING: shakti-vpn.fmp.com: selfcheck request failed: timeout waiting for ACK
Client check: 1 host checked in 30.087 seconds, 1 problem found
On the client, amandad is trying to communicate with the server. From
the debug output in /var/spool/amanda/tmp/amandad:
amandad: time 0.001: security_handleinit(handle=0x50ced0, driver=0x2adfea6ac980
(BSD))
amandad: time 0.001: accept recv REQ pkt:
<<<<<
SERVICE noop
OPTIONS features=ffffffff9ffeffffffff00;
>>>>>
amandad: time 0.001: creating new service: noop
OPTIONS features=ffffffff9ffeffffffff00;
amandad: time 0.003: sending ACK pkt:
<<<<<
>>>>>
amandad: time 0.003: dgram_send_addr(addr=0x50cf10, dgram=0x2adfea6aff68)
amandad: time 0.003: (sockaddr_in *)0x50cf10 = { 2, 739, 10.8.0.4 }
amandad: time 0.003: dgram_send_addr: 0x2adfea6aff68->socket = 0
amandad: time 0.007: sending REP pkt:
<<<<<
OPTIONS features=ffffffff9ffeffffffff00;
>>>>>
amandad: time 0.007: dgram_send_addr(addr=0x50cf10, dgram=0x2adfea6aff68)
amandad: time 0.007: (sockaddr_in *)0x50cf10 = { 2, 739, 10.8.0.4 }
amandad: time 0.007: dgram_send_addr: 0x2adfea6aff68->socket = 0
amandad: time 10.005: timeout
amandad: time 10.005: sending REP pkt:
<<<<<
OPTIONS features=ffffffff9ffeffffffff00;
>>>>>
amandad: time 10.005: dgram_send_addr(addr=0x50cf10, dgram=0x2adfea6aff68)
amandad: time 10.005: (sockaddr_in *)0x50cf10 = { 2, 739, 10.8.0.4 }
amandad: time 10.006: dgram_send_addr: 0x2adfea6aff68->socket = 0
amandad: time 10.029: dgram_recv(dgram=0x2adfea6aff68, timeout=0,
fromaddr=0x2adfea6bff60)
amandad: time 10.029: (sockaddr_in *)0x2adfea6bff60 = { 2, 739, 10.8.0.4 }
amandad: time 10.029: received REQ pkt:
<<<<<
etc ..... until the transaction times out.
tcpdump shows that packets are being received on the server side:
18:29:56.579782 IP shakti-vpn.fmp.com.1025 > vishnu-vpn.fmp.com.739: UDP,
length 50
18:29:56.586474 IP shakti-vpn.fmp.com.1025 > vishnu-vpn.fmp.com.739: UDP,
length 91
18:30:06.583069 IP shakti-vpn.fmp.com.1025 > vishnu-vpn.fmp.com.739: UDP,
length 91
18:30:06.606557 IP shakti-vpn.fmp.com.1025 > vishnu-vpn.fmp.com.739: UDP,
length 91
18:30:16.610980 IP shakti-vpn.fmp.com.1025 > vishnu-vpn.fmp.com.739: UDP,
length 91
18:30:16.635373 IP shakti-vpn.fmp.com.1025 > vishnu-vpn.fmp.com.739: UDP,
length 91
No debug directory for the client is generated on the server, and hence
no debug data for the problem client under ~amanda/tmp/client, although
there is one for the server itself which acts (successfully) as a client
and backs itself up. One other client on the network running a much
older version of amanda has no problems with amcheck, it seems.
If I backversion Amanda to 2.4.5 and build it against an old version of
tar then things work better, and I can connect the the amanda server
with the problem client with no problems, but there are a number of
reasons I'd like to upgrade it since this version is several years old.
Is there perhaps an interface directive in newer versions of amanda
which tells various components which interfaces to listen to?
I'm at a dead end on trying to troubleshoot this. Any advice or help
will be appreciated.
Lindsay Haisley | "Everything works | PGP public key
FMP Computer Services | if you let it" | available at
512-259-1190 | (The Roadie) | http://pubkeys.fmp.com
http://www.fmp.com | |
|
