Hi, I was using kernel-2.6.18-164.15.1 with no known issues. Then, I updated to: Linux 2.6.18-194.17.1.el5 #1 SMP Wed Sep 29 12:50:31 EDT 2010 x86_64 x86_64 x86_64 GNU/Linux I'm aware that there are later updates that patch couple more security bugs. Here is an issue I run with 2.6.18-194.17.1 and am asking if anyone hits that kernel bug. Here are the symptoms: 1) kswapd1 running at 100%, there are not much paging 2) when i logged in, root can't kill -9 ANY processes 3) memory is used up, swap is probably only used up for couple hundred MBs at most (Swap space has couple Gig available) I saw in the logs: kernel: INFO: task sadc:23936 blocked for more than 120 seconds. kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. Annoyingly, I failed to reproduce the issues. I tried to reproduce it with the "stress" software package, using up CPU, memory, etc. But I failed to reproduce those. Reading later release of kernel did not mention this bug. If anyone can contribute to what is happening here, that would be great. Thanks
Hi, (2010/11/13 6:25), My LinuxHAList wrote:> I saw in the logs: > kernel: INFO: task sadc:23936 blocked for more than 120 seconds. > kernel: "echo 0> /proc/sys/kernel/hung_task_timeout_secs" disables > this message.This message is useless to solve problem without Linux developer. Every high load and small trouble machine represents this message. It seems hard disk trouble or memory error. If your disk has write-error happened, retry sequences will be run at, firmware(in HDD), physical driver(in Linux), Logical driver(in Linux scsi) and swapd (in application). Over of 120sec of total these retry sequences, the kernel prints this message in console. you have to change some hardware (hard-disk, memory, cpu, mother-board.) HDD errors and memory errors sometimes be fixed in automatically, You can't reproduce this errors again.> Reading later release of kernel did not mention this bug.Upstream/CentOS community will not release this software fix. :-) Tsuyoshi