>>>>> On Tue, 12 Feb 2019 14:12:08 -0500, mark bergman said:
>
> We run the bacula-dir and bacula-sd (v9.4.1) on a CentOS6 server
> with multiple network interfaces. I'm seeing a problem with a new client
> that I believe is related to asymetric routing and maybe the use of the
> server's IP to generate the MD5 CRAM digest.
FWIW, the MD5 CRAM algorithm is not affected by the IP addresses.
> Here's a simplified network diagram, best viewed with a fixed-width font:
>
>
>
>
> [client 1] [client 2]
> | |
> | |
> | +-----------------------+
> | | |
> {192.168.1.0/24} {172.16.1.0/24} |
> | | |
> | | |
> ________________________________ |
> | 192.168.1.159 172.16.1.159 | |
> | | |
> | bacula-dir & bacula-sd server | |
> | | |
> | 10.20.0.159 | |
> +-------------------------------+ |
> | |
> | |
> {10.20.0.0/16} |
> | |
> +---------------+ {172.16.1.0/24}
> | | |
> _______|______ ___|__________ |
> |10.20.0.10 | | default GW | |
> | new client | | NAT server |------+
> +------------+ +-------------+
>
>
>
> Bacula is able to backup from legacy client1 and client2 machines with
> no problem.
>
> The client bacula-fd.conf files are identical on each client, except
> for the Name field for the FileDaemon itself.
>
> On the bacula server, 10.20.0.159 is a virtual (secondary) IP.
>
> I've added a new client, on a new network (10.20.0.0/16). The new client
> is single-homed, with a default route via a server that does NAT for
> outbound traffic.
>
> The bacula server is able to communicate with new_client via the 10.20.0.0
> network.
>
> The new_client was initially able to communicate with the bacula server
> through the default gateway and NAT to the bacula server at 172.16.1.159.
>
> I have created iptables DNAT rules to rewrite outbound traffic from the
> new_client that is destined for the bacula server on 172.16.1.159 so
> that traffic goes to 10.20.0.159.
>
> Now, new_client sends all traffic to the bacula server server via 10.20.0.0.
>
> I haven't looked at the source code, but I'm wondering if the problem is
> not in the network traffic itself, but in the authentication challenge. Is
> the challenge based on the bacula-dir server's hostname or IP? It appears
> as if the bacula server identifies itself as 172.16.1.159 but the client
> sees traffic from the server on 10.20.0.159.
>
> Here are the server & client debugging outputs:
>
> ------------- Server -----------------
> 1 [root@bacula-server ~]# /opt/bacula/bin/bconsole
> 2 Connecting to Director bacula-server:9101
> 3 bacula-server: bsock.c:851-0 socket=8 who=client
> host=10.20.0.159 port=63069
> 4 bacula-server: job.c:1767-0 wstorage=neoxl80
> 5 bacula-server: job.c:1776-0 wstore=neoxl80 where=Pool resource
> 6 bacula-server: job.c:1430-0 JobId=0 created
> Job=-Console-.2019-02-11_18.42.49_02
> 7 bacula-server: cram-md5.c:69-0 send: auth cram-md5 challenge
> <757376812.1549928569@bacula-server> ssl=0
> 8 bacula-server: cram-md5.c:133-0 cram-get received: auth
> cram-md5 <2004139279.1549928569@bconsole> ssl=0
> 9 bacula-server: cram-md5.c:157-0 sending resp to challenge:
> nH+CS8krrl+Qg9+3F/+TZA
> 10 1000 OK: 103 bacula-server Version: 9.4.1 (20 December 2018)
> 11 Enter a period to cancel a command.
> 12 *estimate
> 13 Select Job resource (1-72): 9
> 14 bacula-server: ua_cmds.c:2608-0 UA Open database
> 15 Using Catalog "MyCatalog"
> 16 bacula-server: job.c:1298-0 Created Client new_client record 56
> 17 bacula-server: fd_cmds.c:314-0 Level=F last start time= job=
> 18 Connecting to Client new_client at 10.20.20.1:9102
> 19 bacula-server: bsockcore.c:299-0 Current 10.20.20.1:9102 All
> 10.20.20.1:9102
> 20 bacula-server: bsockcore.c:228-0 who=Client: new_client
> host=10.20.20.1 port=9102
> 21 bacula-server: bsockcore.c:411-0 OK connected to server
> Client: new_client 10.20.20.1:9102.
According to these debug messages, the server is connecting to client IP
10.20.20.1, not 10.20.0.10.
> 22 bacula-server: fd_cmds.c:116-0 Opened connection with File
> daemon
> 23 bacula-server: authenticate.c:215-0 Sent: Hello Director
> bacula-server calling 103
> 24 bacula-server: authenticate.c:237-0 cram_get_auth failed for
> Client: new_client
> 25 bacula-server: authenticate.c:241-0 Director and File daemon
> passwords or names not the same.
> 26 bacula-server: fd_cmds.c:123-0 Authentication error with FD.
> 27 Failed to connect to Client.
> 28 You have messages.
> 29 *
> --------------------------------------
>
>
>
> ------------- Client -----------------
> 1 bacula-fd: address_conf.c:274-0 Initaddr 0.0.0.0:9102
> 2 new_client: jcr.c:131-0 read_last_jobs seek to 192
> 3 new_client: jcr.c:138-0 Read num_items=0
> 4 new_client: plugins.c:97-0 load_plugins
> 5 new_client: plugins.c:136-0 Found plugin: name=bpipe-fd.so
> len=11
> 6 new_client: filed.c:270-0 filed: listening on port 9102
> 7 new_client: bnet_server.c:86-0 Addresses 0.0.0.0:9102
> 8 new_client: bsock.c:847-0 socket=4 who=client host=172.16.1.159
> port=22958
> 9 new_client: job.c:317-0 <dird: Hello Director bacula_server
> calling 103
> 10 new_client: job.c:340-0 Executing Hello Dir Hello Director
> bacula_server calling 103 command.
> 11 new_client: cram-md5.c:69-0 send: auth cram-md5 challenge
> <1469168970.1549998494@new_client> ssl=0
> 12 new_client: cram-md5.c:71-0 Send challenge comm error.
> ERR=Connection reset by peer
> 13 new_client: authenticate.c:101-0 cram_auth challenge failed for
> Director client
> 14 new_client: Fatal Error at authenticate.c:105 because:
> 15 Incorrect password given by Director at client.
> 16 new_client: job.c:343-0 Quit command loop. Canceled=0
> 17 new_client: job.c:467-0 Calling term_find_files
> 18 new_client: job.c:470-0 Done with term_find_files
> 19 new_client: job.c:473-0 Done with free_jcr
>
> ------------------------------------------------
>
> In particular, note Line 8, where the bacula server seems to identify
> itself as 172.16.1.159 (the primary interface), even though the traffic
> was sent via the 10.20.0.0 network.
The host IP address in line 8 is the address of the server as returned by
accept(2), i.e. at the TCP/IP level, not within the application data.
I suggest you temporarily remove the DNAT rules from the client, because I
don't see why they are needed for the estimate command.
__Martin
_______________________________________________
Bacula-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-users
