BROKER + CLUSTER - stuck

Zeek
#1

Hello all,

Short version : On bro2.5 I cannot seem to get the cluster working with a broker script. that same script is working on a single instance of bro.

Detailed version :

a.bro has a Broker listener linked to myevent. Once myevent is occurring, the string it received is added to a table named tb. On a pybroker remote machine I send updates to the listener.

This scenario works with a single bro instance.

However when I try to run broctl deploy, its stuck on “checking configurations …” and never finish executing.
I do see 7 bro processes that are up (1 manager, 1 proxy + 4 workers).

Any ideas on why is this occurring ?

Also, if I state in a.bro that the Broker should listen to 9999/tcp, what happens in the case of a cluster ? do all instances try to bind to the same address+port ?

How can I send one update from my pybroker script to several bro instances (workers) ?

Thanks in advance

B

#2

For just the hanging issue… Do you perhaps have “exit_only_after_terminate” set in your policy for testing? I’ve seen that hang the config checker.

Dop

#3

Have your pybroker script listen for connections and have all the workers connect to it. Or if you just send updates to the manager but mark the table as &synchronized updates will propagate to the other workers.

#4

Thanks, the exit_only_after_terminate helped

about the &synchronized, only the manager is listening and all nodes are familiar with tb: table[string] of string &create_expire=10min &synchronized.

manager successfully updates tb but it is not synchronized across cluster :

broctl print TEST::tb

manager TEST:tb = {

[“hi”] = [name=“bye”]
}

proxy-1 TEST:tb = {
}
worker-0 TEST:tb = {
}

do I need to invoke something after adding an element to tb, or &synchronized should be taking care of it ?

Thanks

B

#5