On 12/05/2012 04:20 AM, James Bottomley wrote:
On Tue, 2012-12-04 at 16:00 -0600, Chris Friesen wrote:
As another data point, it looks like we may be doing a SEND DIAGNOSTIC
command specifying the default self-test in addition to the background
short self-test. This seems a bit risky and excessive to me, but
apparently the guy that wrote it is no longer with the company.
This is a really bad idea. A lot of disks go out to lunch until the
diagnostics complete (the same goes for SMART diagnostics). This means
that if you do diagnostics on a running device, the drivers start to get
timeouts on commands which are queued waiting for diagnostics to
complete ... if those go over the standard SCSI timeouts, we'll start to
try error recovery and likely have the disaster you see above.
What is the recommended method for monitoring disks on a system that
is likely to go a long time between boots? Do we avoid any in-service
testing and just monitor the SMART data and only test it if something
actually goes wrong? Or should we intentionally drop a disk out of the
array and test it? (The downside of that is that we lose
redundancy since we only have 2 disks.)
What do you mean by "monitoring" ... as in what are you looking for? To
make sure the disk is healthy and responding, a simple test unit ready
works. To look at other parameters, read the mode pages.
Anything that actively causes the disk to go out and check something is
a bad idea in a running environment. Only do this if you can quiesce
the I/O before starting the active diagnostic (or drop the disk from the
array as you suggest).
To be honest, though, modern disks do a whole host of diagnostics as
they write data just to check that it is safely committed, so passive
monitoring should be fine.
James
I don't think that the basic stat gathering (smartctl -a ....) has this kind of
impact, but am worried about the running of the diagnostics,
ric
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
