We are using ESRI 10.5 in an Enterprise environment using SQL Server 2016.
We are seeing gigantic memory creep in the ARcSOC session related to System.CachingTools.GPServerSync. Within 15 minutes of starting it has hit 2.8 gig and after 3 hours it has reached 17 gig (at which point our server turns up its toes). Thread and handle counts are not increasing.
Our current mitigation is to recycle service after 4 hours (I think I will be bringing this down to 1 hour)
Before I start opening an official case with ESRI I was wondering if anyone else has seen this behaviour and can tell me possible causes
datetime | MemUsage(K) | HandleCount | ThreadCount | CPUTime |
13/12/2017 12:15 | 2,831,548 | 1141 | 44 | 2,402,187,500 |
13/12/2017 12:30 | 4,685,008 | 1159 | 46 | 6,837,187,500 |
13/12/2017 12:45 | 6,801,564 | 1050 | 43 | 11,008,906,250 |
13/12/2017 13:00 | 8,299,000 | 1049 | 43 | 15,670,312,500 |
13/12/2017 13:15 | 9,691,264 | 1048 | 43 | 20,743,750,000 |
13/12/2017 13:30 | 10,694,804 | 1046 | 43 | 25,436,562,500 |
13/12/2017 13:45 | 11,255,444 | 1049 | 43 | 30,856,718,750 |
13/12/2017 14:00 | 12,145,916 | 1048 | 43 | 36,877,500,000 |
13/12/2017 14:15 | 13,335,792 | 1049 | 43 | 42,753,437,500 |
13/12/2017 14:30 | 14,612,320 | 1049 | 43 | 48,973,125,000 |
13/12/2017 14:45 | 15,484,756 | 1048 | 43 | 55,534,531,250 |
13/12/2017 15:00 | 16,742,188 | 1048 | 43 | 62,003,281,250 |
13/12/2017 15:15 | 16,830,284 | 1009 | 41 | 66,780,781,250 |