Bug #2362
closedrule reload with workers mode and NFQUEUE not working stable
Description
If you run the IPS mode with NFQUEUE (with --queue-balance) and the workers mode the rule reload with SIGUSR2 might not succeed:
In detect-engine.c packets are injected in every thread to swap to the new ruleset, but this won't happen in runmode workers since there are no incoming queues inq. We then wait for every thread to receive the new rules which could take ages on systems with small amount of traffic since the queue balance from NFQUEUE is not perfectly distributed.
One workaround is to inject packets with iptables but the proper solution would be inject those within the runmode. Any suggestions on how to approach this issue?
Updated by Victor Julien about 7 years ago
I think this could be solved in a similar way as in pfring: by calling TmThreadsCaptureInjectPacket in this case.
Updated by Victor Julien almost 6 years ago
- Status changed from New to Closed
- Assignee changed from OISF Dev to Alexander Gozman
- Target version changed from TBD to 4.1.3
Updated by Victor Julien almost 6 years ago
- Copied to Bug #2824: rule reload with workers mode and NFQUEUE not working stable (4.0.x) added