On Tue, 07 Jun 2005 08:39:50 +1000, Grant Coady wrote:
> Search for your issue in context of conserving battery on laptops
> by reducing hard disk activity. That's the usual group who care.

I'll try this - thanks

> Disk spin-up time to satisfy request may take you over acceptable
> response latencies? Although longer term everything served might
> stay in memory cache automagically for you.

Response times should not be an issue for me, thankfully.

> Then you have the log: -- MARK -- to think about

Hmm. Don't want to be with logging - is there not a way to set the
interval to something longer than 20 min.

On Mon, 06 Jun 2005 22:44:54 -0500, William Hamblen wrote:
> On 2005-06-06, smb <invalid@invalid.invalid> wrote:
>> I have a server than runs 24/7, but much of the time there isn't much
>> demand placed on it. I have set the disk to spin down after 20min of
>> inactivity, but it never seems to reach this threshold. On closer
>> inspection, I see the hard disk light flicks on briefly every 5 secs or
>> so. I can't think what would cause this. Are there any tools that can
>> help my trace what is causing this disk activity?
>
> Something is being written to a log file, probably. Sync runs every
> few seconds and writes to disk the dirty buffers.

I've checked my logs, but nothing is being written as frequenty as every 5
seconds. Not even every 5 min. Once sync has run and written the dirty
buffers to disk, why would it need to do the same 5 seconds later, and
then 5 seconds after that, if no other disk activity is occuring?

On Tue, 07 Jun 2005 13:45:03 -0500, smb <invalid@invalid.invalid> wrote:
> I've checked my logs, but nothing is being written as frequenty as every 5
> seconds. Not even every 5 min. Once sync has run and written the dirty
> buffers to disk, why would it need to do the same 5 seconds later, and
> then 5 seconds after that, if no other disk activity is occuring?

You missed my 'atime' (access time) update looping suggestion?
--Grant

On Tue, 07 Jun 2005 16:19:16 +0100, Claudio Silva wrote:
> In my case, the disks works every 2 secs. I dont know, since it's a
> server, if it has kde or if it's on init 4, but the disk activity stops if
> i turn off the KDE Media manager (Control Center/KDE Components/Service
> Manager). I'm using KDE 3.4.0.

Not running any GUI, runlevel 3.

On Wed, 08 Jun 2005 04:50:37 +1000, Grant Coady wrote:
> On Tue, 07 Jun 2005 13:45:03 -0500, smb <invalid@invalid.invalid> wrote:
>> I've checked my logs, but nothing is being written as frequenty as every 5
>> seconds. Not even every 5 min. Once sync has run and written the dirty
>> buffers to disk, why would it need to do the same 5 seconds later, and
>> then 5 seconds after that, if no other disk activity is occuring?
>
> You missed my 'atime' (access time) update looping suggestion?
> --Grant

Ah yes. I didn't quite get that the first time I read. Have done a bit
of reading and have turned off atime. This might have done the trick -
the hard disk light hasn't flickered for a while now.

Now I've just got to sort out the logging so my disk doesn't have to wake
up to write -- MARK -- in my messages every 20 min.

Thanks.

smb wrote :

> Now I've just got to sort out the logging so my disk doesn't have to wake
> up to write -- MARK -- in my messages every 20 min.

From 'man syslogd':
-m interval
The syslogd logs a mark timestamp regularly. The default interval
between two -- MARK -- lines is 20 minutes. This can be changed with
this option. Setting the interval to zero turns it off entirely.
--
Thomas O.

This area is designed to become quite warm during normal operation.

On Tue, 07 Jun 2005 23:24:48 +0200, Thomas Overgaard wrote:
>
> smb wrote :
>
>> Now I've just got to sort out the logging so my disk doesn't have to wake
>> up to write -- MARK -- in my messages every 20 min.
>
> From 'man syslogd':
> -m interval
> The syslogd logs a mark timestamp regularly. The default interval
> between two -- MARK -- lines is 20 minutes. This can be changed with
> this option. Setting the interval to zero turns it off entirely.

Yeah, I found this last night and modified my rc.syslog. I've not had
any -- MARK -- entries since, so all is well.

Many thanks to all for the responses.