Use Case 8: distributed command/script synchronization, auto-discovery feature

This use case shows extends Use Case 7 and shows as auto-discovery feature works.

Before you can start:

Download and install flom on two different systems: they must be able to contact each other using IP network; it is suggested to play with Use Case 7 before trying this one.

UDP/IP multicast and firewalls

Please pay attention some Linux distros configure a default firewall that silently drop multicast datagrams: disable your firewall or configure it to not drop multicast datagrams.
openSUSE 12.2 is such a distribution.

Activate a flom network daemon and check it’s up & running:

System mojan (IP address 192.168.1.2) will be our flom network daemon (server) system, so we have to activate it with the following commands:

tiian@mojan:/usr$ pgrep flom
tiian@mojan:/usr$ flom -a 192.168.1.2 -A 239.255.0.1 -d -1 -- true
tiian@mojan:/usr$ pgrep flom
2634
tiian@mojan:/usr$ ps -ef | grep flom | grep -v grep
tiian     2634     1  0 21:39 ?        00:00:00 flom -a 192.168.1.2 -A 239.255.0.1 -d -1 -- true
tiian@mojan:/usr$

There was no a running daemon before we started it, now there’s exactly one flom running process with process id 2634.

Check the daemon is serving local requests:

tiian@mojan:/usr$ flom -a 192.168.1.2 -d 0 -- ls
bin  games  include  lib  lib64  local	sbin  share  src
tiian@mojan:/usr$

Switch to the second terminal (second system, IP address 192.168.1.4) and try the same command using the IP address or the network name of flom daemon:

tiian@presanella:/usr$ flom -a 192.168.1.2 -d 0 -- ls
bin  games  include  lib  local  sbin  share  src
tiian@presanella:/usr$ flom -a mojan -d 0 -- ls
bin  games  include  lib  local  sbin  share  src
tiian@presanella:/usr$

From the second terminal try the autodiscovery feature based on UDP/IP multicast:

tiian@presanella:/usr$ flom -A 239.255.0.1 -d 0 -- ls
bin  games  include  lib  local  sbin  share  src
tiian@presanella:/usr$

Experiment this synchronization:

1. inside the first terminal write this command at prompt, but do not press “enter”: “flom -A 239.255.0.1 -d 0 -- ls”
2. inside the second terminal write this command at prompt: “flom -A 239.255.0.1 -d 0 -- sleep 10”
3. now press “enter” key at the second terminal (where you have written “flom -A 239.255.0.1 -d 0 -- sleep 10”)
4. switch to first terminal and press “enter” key

Expected result:

1. the second terminal (system presanella in the above example) pauses for 10 seconds
2. the first terminal (system mojan in the above example) pauses and displays the output of command “ls” after the second terminal sleeping terminates

Terminal 1 output:

tiian@mojan:/usr$ flom -A 239.255.0.1 -d 0 -- ls
bin  games  include  lib  lib64  local	sbin  share  src
tiian@mojan:/usr$

Terminal 2 output:

tiian@presanella:/usr$ flom -A 239.255.0.1 -d 0 -- sleep 10
tiian@presanella:/usr$

Explanation:

command “sleep 10” and command “ls” synchronized: “ls” executed after “sleep 10” completion.
flom command protects (synchronizes) the execution of the command (or script) specified after the – separator on the command line.

Auto-discovery explanation:

IP address 239.255.0.1 is not an address associated to mojan system but a reserved special multicast address. When flom uses the “client” mode specifying “-d 0” it will perform a multicast UDP/IP query, using address 239.255.0.1, to discover if a flom daemon is active and where it is located. The following example shows what happens if the flom daemon is not available:

Terminal 1 output (flom daemon shutdown):

tiian@mojan:/usr$ pgrep flom
2634
tiian@mojan:/usr$ pkill flom
tiian@mojan:/usr$ pgrep flom
tiian@mojan:/usr$

Terminal 2 output (unsuccessful auto-discovery):

tiian@presanella:/usr$ flom -A 239.255.0.1 -d 0 -- sleep 10
flom_client_connect: ret_cod=-104 (ERROR: 'connect' function returned an error condition)
tiian@presanella:/usr$

more details can be retrieved activating tracing:

Terminal 2 output (unsuccessful auto-discovery/tracing):

tiian@presanella:/usr$ export FLOM_TRACE_MASK=0x8
tiian@presanella:/usr$ flom -A 239.255.0.1 -d 0 -- sleep 10
2014-03-09 21:55:39.892322 [7148/0x15c2e00] flom_client_connect
2014-03-09 21:55:39.892512 [7148/0x15c2e00] flom_client_discover_udp
2014-03-09 21:55:39.892549 [7148/0x15c2e00] flom_client_discover_udp: using address '239.255.0.1' and port 28015
2014-03-09 21:55:39.892612 [7148/0x15c2e00] flom_client_discover_udp/getaddrinfo(): [ai_flags=2,ai_family=2,ai_socktype=2,ai_protocol=17,ai_addrlen=16,ai_canonname='239.255.0.1'] 
2014-03-09 21:55:39.892705 [7148/0x15c2e00] flom_client_discover_udp: ai_addr 02 00 6d 6f ef ff 00 01 00 00 00 00 00 00 00 00 
2014-03-09 21:55:39.893078 [7148/0x15c2e00] flom_client_discover_udp: sending discover message...
2014-03-09 21:55:40.891936 [7148/0x15c2e00] flom_client_discover_udp: no answer from UDP/IP multicast discovery
2014-03-09 21:55:40.892077 [7148/0x15c2e00] flom_client_discover_udp/excp=5/ret_cod=4/errno=11 ('Resource temporarily unavailable')
2014-03-09 21:55:40.892111 [7148/0x15c2e00] flom_client_connect: connection failed, a new daemon can not be started because daemon lifespan is 0
2014-03-09 21:55:40.892127 [7148/0x15c2e00] flom_client_connect/excp=4/ret_cod=-104/errno=11
flom_client_connect: ret_cod=-104 (ERROR: 'connect' function returned an error condition)
tiian@presanella:/usr$

Extra considerations related to -d option:

-d (--daemon-lifespan) is the parameter used to specify how long a flom daemon must run.
-d -1 is used in the above example with the meaning “start a daemon and don’t stop it”
-d 0 is used in the above example with the meaning “don’t start a daemon”
First usage is typically server oriented, while the second usage is typically client oriented.
If you are not happy to specify “-d 0” for every client command, you can set Lifespan option in a configuration file.

To terminate a running flom daemon, simply use kill or pkill command:

tiian@mojan:/usr$ pgrep flom
2198
tiian@mojan:/usr$ pkill flom
tiian@mojan:/usr$ pgrep flom
tiian@mojan:/usr$

Summary

This example of command flom allows you to synchronize commands/scripts running on different systems and you don’t have to start flom daemon on a specific system: it may be started on any system that can answer to UDP/IP multicast queries.
With auto-discovery feature all the information needed to reach the flom daemon is the UDP/IP multicast address and port (multicast group): this can save you a lot of configuration pains.