mirror of
https://github.com/zeek/zeek.git
synced 2025-10-02 14:48:21 +00:00
Merge remote-tracking branch 'origin/topic/arne.welzel/fix-poll-interval-is-live'
* origin/topic/arne.welzel/fix-poll-interval-is-live: iosource/Manager: Fix poll_interval updating using not-yet valid IsLive()
This commit is contained in:
commit
6ac72a31bf
4 changed files with 37 additions and 3 deletions
26
CHANGES
26
CHANGES
|
@ -1,3 +1,29 @@
|
|||
6.0.0-dev.353 | 2023-04-12 13:23:29 -0700
|
||||
|
||||
* iosource/Manager: Fix poll_interval updating using not-yet valid IsLive() (Arne Welzel, Corelight)
|
||||
|
||||
Testing io_poll_interval_live tweaks with @dopheide-esnet on a Myricom based
|
||||
system to reduce CPU usage showed no visible effect.
|
||||
|
||||
It turns out, the pkt_src->IsLive() call used to update poll_interval is only
|
||||
valid *after* calling ->Register() with the source. The conditional updating
|
||||
of the poll_interval introduced in 4fa3e4b9b4d78e6f20f42b9dbd85216403be07db
|
||||
never worked out how it was intended to.
|
||||
|
||||
The fix ensures that
|
||||
|
||||
* we actually use a poll_interval of 10 in the live case
|
||||
* changing io_poll_interval_live does have an effect
|
||||
|
||||
This is a bit of a major change due to lowering the default poll_interval
|
||||
by a magnitude, but that seemed to have been the intention always. It's also
|
||||
tunable via redef, so worst case it can be adapted via configuration.
|
||||
|
||||
As reference, with the default a Pcap::non_fd_timeout of 20usec *and* a
|
||||
poll_interval of 100, theoretically we'd be trying to ask a non-selectable
|
||||
packet source 500000 per second for a new packet. This is not a likely packet
|
||||
rate that a single worker would currently observe or manage to process.
|
||||
|
||||
6.0.0-dev.350 | 2023-04-11 15:41:31 -0700
|
||||
|
||||
* Only use -Wno-changes-meaning if it exists (Tim Wojtulewicz, Corelight)
|
||||
|
|
4
NEWS
4
NEWS
|
@ -191,6 +191,10 @@ Changed Functionality
|
|||
processing input. Specifically in zeek -r scenarios, that is not the wanted
|
||||
behaviour.
|
||||
|
||||
- The IO loop's poll interval is now correctly reduced from 100 to 10 for
|
||||
live packet sources. This should lower CPU usage for deployments with
|
||||
non-selectable packet sources.
|
||||
|
||||
Removed Functionality
|
||||
---------------------
|
||||
|
||||
|
|
2
VERSION
2
VERSION
|
@ -1 +1 @@
|
|||
6.0.0-dev.350
|
||||
6.0.0-dev.353
|
||||
|
|
|
@ -402,12 +402,16 @@ void Manager::Register(PktSrc* src)
|
|||
{
|
||||
pkt_src = src;
|
||||
|
||||
Register(src, false);
|
||||
|
||||
// Once we know if the source is live or not, adapt the
|
||||
// poll_interval accordingly.
|
||||
//
|
||||
// Note that src->IsLive() is only valid after calling Register().
|
||||
if ( src->IsLive() )
|
||||
poll_interval = BifConst::io_poll_interval_live;
|
||||
else if ( run_state::pseudo_realtime )
|
||||
poll_interval = 1;
|
||||
|
||||
Register(src, false);
|
||||
}
|
||||
|
||||
static std::pair<std::string, std::string> split_prefix(std::string path)
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue