[Bro] Memory Issue with Bro

[Azoff, Justin S]

Well that doesn't look great, but could be a lot worse.  Hard to say without knowing what it looked like before the patch.

The fact that pending ever goes down at all is a good sign, but pending=0 is really the optimal state.

-- 
- Justin Azoff

> On Oct 23, 2015, at 9:21 AM, Mike Waite <mfw113 at psu.edu> wrote:
> 
> Patch applied, after 15 minutes I am seeing
> 
> Oct 23 09:00:43	manager	child	-	-	-	info	selects=300000 canwrites=216206 pending=0
> Oct 23 09:01:29	manager	child	-	-	-	info	selects=400000 canwrites=216206 pending=0
> Oct 23 09:02:08	manager	child	-	-	-	info	selects=500000 canwrites=216552 pending=0
> Oct 23 09:02:49	manager	child	-	-	-	info	selects=600000 canwrites=216557 pending=0
> Oct 23 09:03:34	manager	child	-	-	-	info	selects=700000 canwrites=216557 pending=0
> Oct 23 09:04:29	manager	child	-	-	-	info	selects=800000 canwrites=255305 pending=4007
> Oct 23 09:05:21	manager	child	-	-	-	info	selects=900000 canwrites=355305 pending=6593
> Oct 23 09:06:13	manager	child	-	-	-	info	selects=1000000 canwrites=455305 pending=6003
> Oct 23 09:07:04	manager	child	-	-	-	info	selects=1100000 canwrites=555305 pending=3077
> Oct 23 09:07:55	manager	child	-	-	-	info	selects=1200000 canwrites=640438 pending=3399
> Oct 23 09:08:45	manager	child	-	-	-	info	selects=1300000 canwrites=740438 pending=3163
> Oct 23 09:09:36	manager	child	-	-	-	info	selects=1400000 canwrites=840438 pending=5245
> Oct 23 09:10:25	manager	child	-	-	-	info	selects=1500000 canwrites=940438 pending=6027
> Oct 23 09:11:15	manager	child	-	-	-	info	selects=1600000 canwrites=1040438 pending=6713
> Oct 23 09:12:01	manager	child	-	-	-	info	selects=1700000 canwrites=1140438 pending=5713
> Oct 23 09:12:50	manager	child	-	-	-	info	selects=1800000 canwrites=1240438 pending=6747
> Oct 23 09:13:39	manager	child	-	-	-	info	selects=1900000 canwrites=1340438 pending=7417
> Oct 23 09:14:32	manager	child	-	-	-	info	selects=2000000 canwrites=1440438 pending=13117
> Oct 23 09:15:10	manager	child	-	-	-	info	selects=2100000 canwrites=1540438 pending=20825
> Oct 23 09:15:59	manager	child	-	-	-	info	selects=2200000 canwrites=1640438 pending=18539
> Oct 23 09:16:47	manager	child	-	-	-	info	selects=2300000 canwrites=1740438 pending=15881
> Oct 23 09:17:35	manager	child	-	-	-	info	selects=2400000 canwrites=1840438 pending=15389
> Oct 23 09:18:28	manager	child	-	-	-	info	selects=2500000 canwrites=1940438 pending=16685
> Oct 23 09:19:18	manager	child	-	-	-	info	selects=2600000 canwrites=2040438 pending=17031
> 
> 
> I will let you know about the mem usage after a bit
> 
> --
> Mike Waite
> CyberSecurity Intrusion Analyst
> Office of Information Security
> The Pennsylvania State University
> ↪ 15-10-22 10:22:18, Azoff, Justin S <jazoff at illinois.edu>:
>>> On Oct 22, 2015, at 8:12 AM, Mike Waite <mfw113 at psu.edu> wrote:
>>> 
>>> I know we are still seeing issues with the manager child proccess.  The process will consume over 200GB of RAM in 8 hours.
>>> 
>> 
>> Give the attached patch a try.
>> 
>> 
>> 
>> Monitor by using
>> 
>>   cat logs/current/communication.log |egrep 'manager.child'
>> 
>> And check to see if pending=0 or at least not growing.
>> 
>> 
>> --
>> - Justin Azoff
>> 
>> 
> 
>