This is 2.4.5 with Andrea Arcangeli's aa1 patch compiled with himem: Why is kswapd using so much CPU? If you reboot the machine and run the same user process kswapd CPU usage is almost 0% and none of the swap is used. This machine was upgraded from 2.2 and we did not have the luxury of re-partitioning it support the "new" 2.4 swap size requirements. After running for a few days with relatively constant memory usage: vmstat: procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 2 0 1 136512 5408 504 209744 0 0 0 2 19 49 10 26 64 top: 5:38pm up 3 days, 19:44, 2 users, load average: 2.08, 2.13, 2.15 34 processes: 32 sleeping, 2 running, 0 zombie, 0 stopped CPU0 states: 16.0% user, 56.4% system, 16.2% nice, 26.3% idle CPU1 states: 11.1% user, 57.0% system, 11.0% nice, 31.3% idle Mem: 1028804K av, 1023744K used, 5060K free, 0K shrd, 504K buff Swap: 136512K av, 136512K used, 0K free 209876K cached PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND 28442 root 18 10 898M 812M 36188 R N 56.0 80.9 296:12 gateway.smart.5 28438 root 16 10 898M 811M 35084 S N 43.7 80.7 291:03 gateway.smart.5 5 root 9 0 0 0 0 SW 37.6 0.0 164:58 kswapd 2509 root 18 0 492 492 300 R 2.5 0.0 0:00 top 1 root 9 0 68 0 0 SW 0.0 0.0 0:08 init 2 root 9 0 0 0 0 SW 0.0 0.0 0:00 keventd 3 root 19 19 0 0 0 SWN 0.0 0.0 1:11 ksoftirqd_CPU0 4 root 19 19 0 0 0 SWN 0.0 0.0 1:04 ksoftirqd_CPU1 6 root 9 0 0 0 0 SW 0.0 0.0 0:00 kreclaimd 7 root 9 0 0 0 0 SW 0.0 0.0 0:00 bdflush 8 root 9 0 0 0 0 SW 0.0 0.0 0:07 kupdated 11 root 9 0 0 0 0 SW 0.0 0.0 0:00 scsi_eh_0 315 root 9 0 100 0 0 SW 0.0 0.0 0:00 syslogd Hope this helps - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/