Hello all,
As of recently we have been getting really high loads on 1 or 2 servers sporadically and I am trying to figure out if its related to our concurrency limits being set to high. Here is a excerpt from my prstat when the loads go really high:
PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP 5288 pmx 149M 119M run 46 0 0:00:13 2.3% pmx-milter/2 5301 pmx 149M 119M run 47 0 0:00:13 2.3% pmx-milter/2 5344 pmx 145M 109M run 38 0 0:00:10 2.2% pmx-milter/2 5338 pmx 145M 110M run 46 0 0:00:10 2.2% pmx-milter/2 5307 pmx 149M 119M run 39 0 0:00:13 2.2% pmx-milter/2 5292 pmx 149M 119M run 48 0 0:00:13 2.2% pmx-milter/2 5325 pmx 145M 110M run 48 0 0:00:10 2.2% pmx-milter/2 5348 pmx 145M 107M run 48 0 0:00:09 2.2% pmx-milter/2 5295 pmx 149M 119M run 45 0 0:00:13 2.2% pmx-milter/2 5289 pmx 145M 115M run 48 0 0:00:11 2.2% pmx-milter/2 5304 pmx 149M 119M run 44 0 0:00:13 2.2% pmx-milter/2 5305 pmx 149M 119M run 45 0 0:00:13 2.1% pmx-milter/2 5298 pmx 149M 119M run 47 0 0:00:13 2.1% pmx-milter/2 5309 pmx 149M 119M run 48 0 0:00:12 2.1% pmx-milter/2 5334 pmx 145M 105M cpu0 55 0 0:00:09 2.1% pmx-milter/2 5306 pmx 149M 119M run 54 0 0:00:13 2.0% pmx-milter/2 5294 pmx 145M 113M run 39 0 0:00:10 2.0% pmx-milter/2 5332 pmx 145M 109M run 44 0 0:00:10 2.0% pmx-milter/2 5327 pmx 145M 105M run 54 0 0:00:09 2.0% pmx-milter/2 5336 pmx 145M 102M run 54 0 0:00:08 2.0% pmx-milter/2 5346 pmx 145M 108M run 47 0 0:00:09 2.0% pmx-milter/2 5323 pmx 145M 107M run 49 0 0:00:09 2.0% pmx-milter/2 5287 pmx 149M 119M run 54 0 0:00:12 2.0% pmx-milter/2 5293 pmx 149M 118M run 54 0 0:00:12 2.0% pmx-milter/2 5296 pmx 149M 119M run 48 0 0:00:12 2.0% pmx-milter/2 5318 pmx 145M 106M run 46 0 0:00:09 1.9% pmx-milter/2 4906 pmx 151M 140M run 48 0 0:00:41 1.9% pmx-milter/2 5354 pmx 145M 107M run 49 0 0:00:09 1.9% pmx-milter/2 5320 pmx 145M 102M run 44 0 0:00:08 1.9% pmx-milter/2 5285 pmx 149M 119M run 46 0 0:00:13 1.9% pmx-milter/2 5290 pmx 149M 119M run 45 0 0:00:13 1.9% pmx-milter/2 5330 pmx 145M 98M run 38 0 0:00:08 1.9% pmx-milter/2 5291 pmx 149M 119M run 49 0 0:00:13 1.9% pmx-milter/2 5299 pmx 149M 119M run 45 0 0:00:12 1.8% pmx-milter/2 4872 pmx 152M 140M run 45 0 0:00:38 1.8% pmx-milter/2 5286 pmx 149M 119M run 49 0 0:00:14 1.8% pmx-milter/2 5341 pmx 145M 108M run 39 0 0:00:09 1.8% pmx-milter/2 5316 pmx 145M 109M run 49 0 0:00:10 1.8% pmx-milter/2 5352 pmx 145M 107M run 39 0 0:00:09 1.8% pmx-milter/2 5343 pmx 145M 107M run 49 0 0:00:09 1.7% pmx-milter/2 5308 pmx 145M 115M run 49 0 0:00:11 1.7% pmx-milter/2 4898 pmx 152M 141M run 48 0 0:00:36 1.5% pmx-milter/2 5350 pmx 145M 102M run 44 0 0:00:08 1.5% pmx-milter/2 5371 pmx 24M 23M run 54 0 0:00:08 1.5% pmx-reports-con/1 4909 pmx 155M 144M run 47 0 0:01:10 1.5% pmx-milter/2 4911 pmx 173M 163M run 45 0 0:02:40 1.4% pmx-milter/2 4907 pmx 153M 142M run 46 0 0:00:51 1.3% pmx-milter/2 5415 pmx 15M 14M run 45 0 0:00:00 1.2% ppm/1 4900 pmx 152M 140M sleep 59 0 0:00:37 0.9% pmx-milter/2 4903 pmx 152M 141M sleep 59 0 0:00:43 0.8% pmx-milter/2 4901 pmx 152M 142M sleep 59 0 0:00:40 0.7% pmx-milter/2 4887 pmx 151M 141M sleep 59 0 0:00:34 0.5% pmx-milter/2 5328 postfix 2856K 2056K sleep 59 0 0:00:00 0.3% cleanup/1 4730 postfix 3856K 2584K sleep 59 0 0:00:00 0.3% smtpd/1 Total: 279 processes, 688 lwps, load averages: 51.07, 33.42, 21.40
I am thinking the loads are caused by an increase in spam, but what can we do to keep the system from getting pegged by the milter process? Should we lower the concurrency? Our concurrency is set to 50 as of the present. The system is a Sunfire v480 with 8GB ram, with 2 1.2 sparc cpu's.
Thanks
This thread was automatically locked due to age.
