Dirk.Laurenz@fujitsu-siemens.com
2005-Sep-22 11:49 UTC
[Samba] High CPU Time an Load Avarage on our Samba Server
Hello list, how could this happen? The Server doesn't respond from time to time with a high load avarage. We found a suspicious smbd process: top - 13:43:07 up 1 day, 2:27, 5 users, load average: 32.49, 58.41, 37.95 Tasks: 1196 total, 5 running, 1190 sleeping, 0 stopped, 1 zombie Cpu0 : 14.7% us, 3.8% sy, 0.0% ni, 79.8% id, 1.3% wa, 0.0% hi, 0.3% si Cpu1 : 1.3% us, 84.6% sy, 0.0% ni, 14.1% id, 0.0% wa, 0.0% hi, 0.0% si Cpu2 : 15.0% us, 6.4% sy, 0.0% ni, 76.7% id, 0.6% wa, 0.0% hi, 1.3% si Cpu3 : 10.9% us, 16.0% sy, 0.0% ni, 72.8% id, 0.0% wa, 0.0% hi, 0.3% si Mem: 6231672k total, 6105452k used, 126220k free, 12944k buffers Swap: 8418016k total, 232k used, 8417784k free, 1636920k cached PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ GROUP COMMAND 19256 PREUSSAG 25 0 14804 8168 3520 R 99.9 0.1 116:00.06 PREUSSAG smbd This is a really big file server: agestt71:~ # ps -ef | grep smbd |wc -l 1014 here's an sar -A: 10:15:01 runq-sz plist-sz ldavg-1 ldavg-5 ldavg-15 10:18:01 2 1262 2.17 2.16 1.82 10:21:01 1 1264 2.08 2.16 1.87 10:24:01 1 1269 2.09 2.18 1.93 10:27:01 1 1262 2.20 2.14 1.95 10:30:01 1 1262 2.01 2.15 1.99 10:33:01 1 1265 2.37 2.20 2.02 10:36:01 1 1264 2.33 2.26 2.08 10:39:01 1 1268 2.32 2.29 2.11 10:42:20 31 1289 340.91 128.06 48.26 10:45:01 1 1276 37.48 92.10 48.24 10:48:01 1 1283 4.44 51.58 40.18 10:51:01 0 1275 1.69 29.00 33.40 10:54:01 0 1272 1.42 16.54 27.77 10:57:01 1 1271 1.35 9.67 23.11 11:00:01 0 1269 1.50 5.94 19.28 11:03:01 1 1272 1.38 3.87 16.12 11:06:01 2 1267 1.29 2.70 13.50 11:09:29 79 1297 190.52 83.20 41.67 11:12:01 0 1269 18.08 52.06 36.16 11:15:01 0 1274 2.36 29.17 30.05 11:18:01 1 1270 1.53 16.63 25.00 11:22:25 50 1296 191.30 89.21 49.85 11:24:01 2 1293 41.52 65.89 45.36 11:24:01 runq-sz plist-sz ldavg-1 ldavg-5 ldavg-15 11:28:02 141 1308 129.16 68.58 48.71 11:30:01 1 1297 115.12 98.49 63.00 11:33:01 0 1303 6.91 54.53 52.14 11:37:13 50 1302 234.47 136.50 83.73 11:39:01 1 1305 41.48 96.57 75.20 11:42:01 1 1307 5.10 54.24 62.47 11:45:01 3 1311 2.61 30.84 51.91 11:48:01 1 1309 2.31 17.92 43.16 11:51:01 2 1288 2.12 10.78 35.93 11:54:01 2 1273 2.35 7.01 30.02 11:57:01 1 1266 2.20 4.84 25.12 12:00:01 3 1272 2.46 3.72 21.10 12:03:01 1 1277 2.19 3.03 17.77 12:06:02 3 1279 2.45 2.72 15.04 12:09:02 1 1270 2.68 2.64 12.83 12:12:01 1 1256 2.12 2.40 10.94 12:15:01 4 1251 2.50 2.48 9.46 12:18:01 1 1251 2.09 2.28 8.15 12:21:01 1 1248 2.09 2.23 7.09 12:24:01 1 1253 2.48 2.31 6.26 12:27:01 1 1246 2.26 2.35 5.58 12:30:01 4 1249 2.49 2.36 5.00 12:33:11 173 1249 190.71 79.53 33.47 12:33:11 runq-sz plist-sz ldavg-1 ldavg-5 ldavg-15 12:36:01 1 1253 16.15 51.05 30.60 12:39:01 1 1251 3.20 29.10 25.65 12:42:01 1 1254 2.36 16.97 21.53 12:45:01 1 1250 2.31 10.35 18.14 12:48:01 1 1253 2.41 6.84 15.40 12:51:01 1 1247 2.16 4.70 13.06 12:54:01 2 1250 2.28 3.63 11.16 12:57:01 1 1240 2.29 3.03 9.60 13:00:01 1 1238 2.34 2.70 8.31 13:03:01 1 1243 2.68 2.65 7.29 13:06:01 2 1240 2.23 2.47 6.40 13:09:01 1 1232 2.30 2.39 5.67 13:12:01 1 1237 2.63 2.50 5.13 13:15:01 2 1236 2.51 2.46 4.64 13:18:01 1 1244 2.35 2.47 4.26 13:21:01 1 1237 2.70 2.52 3.95 13:24:01 3 1230 2.14 2.36 3.64 13:27:01 2 1224 2.55 2.45 3.44 13:32:38 30 1240 143.09 65.32 27.01 13:33:01 2 1221 103.45 61.32 26.51 13:36:01 2 1216 8.40 34.90 22.34 13:41:14 146 1228 156.22 75.74 39.82 13:42:01 1 1203 98.72 73.12 40.72 13:42:01 runq-sz plist-sz ldavg-1 ldavg-5 ldavg-15 13:45:01 2 1203 7.22 41.16 33.98 Average: 4 854 12.59 10.92 9.50 As you can see, the load avarage ist from time to time up more tha 300 high. Mit freundlichem Gru?, Dirk Laurenz Systems Engineer PSO - Professional Service Organisation Fujitsu Siemens Computers Hildesheimer Strasse 25 30880 Laatzen Germany Telephone: +49 (511) 84 89 - 18 08 Telefax: +49 (511) 84 89 - 25 18 08 Mobile: +49 (170) 22 10 781 Email: mailto:dirk.laurenz@fujitsu-siemens.com Internet: http://www.fujitsu-siemens.com http://www.fujitsu-siemens.de/rl/servicesupport/itdienstleistungen/competencecenter.html *******************************************************************************************************************