| 1 |
Yo Mark! |
| 2 |
|
| 3 |
On Fri, 21 Jun 2013 11:38:00 -0700 |
| 4 |
Mark Knecht <markknecht@×××××.com> wrote: |
| 5 |
|
| 6 |
> On the read side I'm not sure if I'm understanding your point. I agree |
| 7 |
> that a so-designed RAID1 system could/might read smaller portions of a |
| 8 |
> larger read from RAID1 drives in parallel, taking some data from one |
| 9 |
> drive and some from another drive, and then only take action |
| 10 |
> corrective if one of the drives had troubles. However I don't know |
| 11 |
> that mdadm-based RAID1 does anything like that. Does it? |
| 12 |
|
| 13 |
It surely does. I have confirmed that at least monthly since md has |
| 14 |
existed in the kernel. |
| 15 |
|
| 16 |
> It seems to me that unless I at least _request_ all data from all |
| 17 |
> drives and minimally compare at least some error flag from the |
| 18 |
> controller telling me one drive had trouble reading a sector then how |
| 19 |
> do I know if anything bad is happening? |
| 20 |
|
| 21 |
Correct. You cant' tell if you can read something without trying to |
| 22 |
read it. Which is why you should do a full raid rebuild every week. |
| 23 |
> |
| 24 |
> Or maybe you're saying it's RAID1 and I don't know if anything bad is |
| 25 |
> happening _unless_ I do a scrub and specifically check all the drives |
| 26 |
> for consistency? |
| 27 |
|
| 28 |
No. A simple read will find the problem. But given it is RAID1 the only |
| 29 |
way to be sure to read from both dirves is a raid rebuild. |
| 30 |
|
| 31 |
> I do mdadm scrubs at least once a week. I still do them by hand. They |
| 32 |
> have never appeared terribly expensive watching top or iotop but |
| 33 |
> sometimes when I'm watching NetFlix or Hulu in a VM I get more pauses |
| 34 |
> when the scrub is taking place, but it's not huge. |
| 35 |
|
| 36 |
Which is why you should cron jothem at oh-dark-thirty. |
| 37 |
> |
| 38 |
> I agree that RAID5 gives you an opportunity to get things fixed, but |
| 39 |
> there are folks who lose a disk in a RAID5, start the rebuild, and |
| 40 |
> then lose a second disk during the rebuild. |
| 41 |
|
| 42 |
Because they failed to do weekly rebuilds. |
| 43 |
|
| 44 |
> Not that I would ever run the array degraded but that I |
| 45 |
> could still tolerate a second loss while the rebuild was happening and |
| 46 |
> hopefully get by. |
| 47 |
|
| 48 |
Sadly most people make their RAID5 or RAID6 out of brand new, |
| 49 |
consecutively serial numbered drives. They then get the exactly the |
| 50 |
same temp, voltage, humidity, seek stress until they all fail within |
| 51 |
days of each other. I have personally seen 4 of 5 drives in a RAID5 |
| 52 |
fail within 3 days many times. Usually on a Friday where the tech |
| 53 |
decides the drive replacement can wait until Monday. |
| 54 |
|
| 55 |
Your only protection against a full RAIDx failure is an offsite backup. |
| 56 |
|
| 57 |
RGDS |
| 58 |
GARY |
| 59 |
--------------------------------------------------------------------------- |
| 60 |
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97701 |
| 61 |
gem@××××××.com Tel:+1(541)382-8588 |
Archives