zfs send/recv invalid data

[Artem Belevich]

On Fri, Mar 8, 2013 at 1:40 PM, Larry Rosenman <ler at lerctr.org> wrote:
> On 2013-03-07 22:47, Larry Rosenman wrote:
>>
>> Will try this with tonights stream(s) if it fails.
>>
>> appreciate.....
>
> [trimmed copied text]
>
> I worked fine last night from cron(8).
>
> Will keep an eye on it.
>
> I hate intermittent / unpredictable failures with a passion :)

Do you use ECC memory? If not, it may be a good idea to test memory on
both machines if you didn't do it already.

If you do use ECC, grep your logs for 'MCA'. If there are correctable
errors, they would typically be reported via machine check exceptions
and logged. YMMV. I was unable to make my Asus board (P5BV/SAS) report
anything at all even though in one experiment I had physically taped
off one data bit on one DIMM -- system still boots and works, but I
get no exceptions. Supermicro board I had at work (X8-something) did
report correctable ECC errors when one DIMM went marginally bad.

--Artem