[collectd] Query: Odd "scsi err" messages to /var/log/messages while collectd is running

[Tim Chipman]

Hi all,

I've been testing out collected on one system for a few months, and just
installed it onto a "pre-production server" I'm working on today. (CentOS
4.2 x86_64, sempron3300+ /64bit with 1gig ram, mirrored 250gb SATA HDDs)

An odd thing: Collectd appears to be working fine, but as long as it is
running, I get these messages thrown to /var/log/messages on approximately 2
minute increments:

Mar 21 16:03:20 docs collectd[31787]: cpufreq found 1 cpu(s)
Mar 21 16:03:53 docs kernel: SCSI error : <0 0 0 0> return code = 0x8000002
Mar 21 16:03:53 docs kernel: Invalid sda: sense key No Sense
Mar 21 16:03:53 docs kernel: Additional sense: Filemark detected

If I stop the collectd daemon, the messages stop piling up.  If I restart
the daemon, the messages resume.  Hence my belief they are related.

I Suspect ? it relates to how Collectd is polling for information from the
drives, possibly the fact that SATA drives are seen to the system as /dev/sd
devices, and somehow? collectd may be assuming they are scsi devices,
treating them with commands that aren't working correctly?  However, I'm not
positive (alas) what this really is.

Any thoughts / comments on how to resolve this would certainly be
appreciated.

Many thanks,

Tim


[ps - I'm running the collectd as per the pre-compiled RPM binary for RHEL 4
64bit, ie, the URL this was obtained from was

http://verplant.org/collectd/files/collectd-3.7.0-1.rhel4.x86_64.rpm ]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.verplant.org/pipermail/collectd/attachments/20060321/2a471c37/attachment.htm