99% cpu usage due to flush-8:0 & flush-8:16

propofol · 07-28-2010, 12:38 AM

I have recently upgraded my Debian unstable to the new custom 2.6.34 kernel. The system is unresponsive and top shows two processes (flush-8:0 & flush-8:16) taking up most of the cpu time.

I would appreciate it if anyone could tell me what these processes does and what I could do to fix the problem.

Regards,
Stefan

syg00 · 07-28-2010, 01:00 AM

I'd be guessing that's the pdflush replacement - have a read of this.
As to what to do - is there any (heavy) I/O load ?.

propofol · 07-29-2010, 12:18 AM

Thanks for the reply. There was no major I/O going on. It seems I must have installed noflushd at some stage. The problem disappeared when I stopped these processes and uninstalled noflushd. I am not sure if this is related to the info about pdflush.

Regards,
Stefan

syg00 · 07-29-2010, 05:48 AM

I would think so - just forget about noflushd.

dbbolton · 09-09-2011, 03:03 PM

I just noticed this same issue in Debian wheezy/sid today (2011-Sep-09), except I also had a huge I/O bottle neck, as reported by iotop. Strangely, "aptitude search ~iflush" shows no results.