Regular CPU hog

[Harald Becker]

 Hi!
>   216 root     /usr/sbin/ntpd
>   231 root     /sbin/syslogd -n -m 0
>   232 root     /sbin/klogd -n
>   233 root     /usr/bin/tail -f /var/log/messages
>   264 root     tail -f /var/log/messages
> There are several processes in that list I'm not familiar with. Are there any 
> that are succeptible to perform regular events like this ?
Some hints that possibly help to find the problem:

- Are there any regular messages going to the syslog?
Look at the timestamps,do they anyhow correlate with your problem?

- To which type of device are those messages written?
RAM is ok, HardDisk may cause slow delays on high message rates, Flash
and other slow devices can definitely cause delays as described, if
there are message writes.

- otherwise try the following:
kill klogd, syslogd and that tail processes
kill ntpd (just in case)

klogd = Send kernel log messages to syslogd
syslogd = Write syslog messages to log file
ntpd = synchronize local clock with network clock server
dropbear = ssh server (allows ssh access to your device)
tail -f = read file as it grows and send text to stdout
processes in [] are kernel threads (can't be killed)

- Does any prosess write data to a slow tty device (JTAG console)?
Massive output on such devices can produce buffer overflows which may
result in described short delays of other processes (temporary out of
kernel memory).

- Does your process contain a memory garbage collector and you are low
on system memory?
General: Is there enough free memory for io buffers?

--
Harald