Multiple instances of watchdog, migration, and ksoftirqd
 

Hi there --

I recently completed a scratch install of Fedora Core 14 64-bit onto a system, and while everything appears to be running properly, I noticed there are almost 300 processes running at a given time on the server.

The top command indicated there were ten-plus instances of the watchdog, migration, and ksoftirqd daemons running on the system. I am not familiar with these services, so I wanted to know what they are, and is there a need to have this many instances of each running on the system? If the answer to the latter question is no, will simply running the kill <pid> command be sufficient to remove the excess processes? Thanks.

Listing processes with 'ps' where a process' argv[0] is in brackets ([watchdog/n], [watchdog/n], [scsi_eh_n] etc, etc) are kernel threads (this has been dealt with in other threads: search LQ?). So unless one or more of those kernel threads consistently contribute to a high system load they don't need investigating. If you want simple performance indicators wrt what bottlenecks userland processes encounter use SAR tools (iostat, vmstat etc), atop, dstat, collectl and such.

@more instance of same process
 

These are per cpu kernel threads. If you 16 CPUs you will get following output (see thread numbers from 0 to 15)

# ps aux | grep migration
root 2 0.0 0.0 0 0 ? S< Feb22 0:27 [migration/0]
root 5 0.0 0.0 0 0 ? S< Feb22 1:07 [migration/1]
root 8 0.0 0.0 0 0 ? S< Feb22 0:38 [migration/2]
root 11 0.0 0.0 0 0 ? S< Feb22 0:22 [migration/3]
root 14 0.0 0.0 0 0 ? S< Feb22 0:23 [migration/4]
root 17 0.0 0.0 0 0 ? S< Feb22 0:06 [migration/5]
root 20 0.0 0.0 0 0 ? S< Feb22 0:01 [migration/6]
root 23 0.0 0.0 0 0 ? S< Feb22 0:01 [migration/7]
root 26 0.0 0.0 0 0 ? S< Feb22 0:04 [migration/8]
root 29 0.0 0.0 0 0 ? S< Feb22 0:29 [migration/9]
root 32 0.0 0.0 0 0 ? S< Feb22 0:19 [migration/10]
root 35 0.0 0.0 0 0 ? S< Feb22 0:12 [migration/11]
root 38 0.0 0.0 0 0 ? S< Feb22 0:09 [migration/12]
root 41 0.0 0.0 0 0 ? S< Feb22 0:19 [migration/13]
root 44 0.0 0.0 0 0 ? S< Feb22 0:16 [migration/14]
root 47 0.0 0.0 0 0 ? S< Feb22 0:09 [migration/15]

You will find similar output for ksoftirqd, watchdog and other per-cpu-kernel threads.

The problem is when you start seeing them in top eating some CPU. Are you?. That actually means you have big load on your server.