DSM-G600, DNS-3xx and NSA-220 Hack Forum
Unfortunately no one can be told what fun_plug is - you have to see it for yourself.
You are not logged in.
Announcement
IRC Channel #funplug on irc.freenode.org
#1 2009-06-15 19:40:27
- oberkeim
- Member
- Registered: 2009-02-23
- Posts: 5
Raid-1 problem
Help me, take a look at the attached pic.
The lights on the device are still blue, I just got skeptic after one of the both HDD lights stopped flashing when accesing the station.
Both drives are working perfectly, when connecting directly to a PC.
Offline
#2 2009-06-15 19:44:44
- oberkeim
- Member
- Registered: 2009-02-23
- Posts: 5
Re: Raid-1 problem
Where is my attachment?
OK: Drives degraded, Ver. 1.06, Auto-Rebuild disabled, images on the web-frontend broken.
I am not sure, if I may use "Manually rebuild now". The station is working fine, but one drive is not active anymore.
Offline
#3 2009-06-16 03:39:51
- fordem
- Member
- Registered: 2007-01-26
- Posts: 1938
Re: Raid-1 problem
Back up your data, click manually rebuild and see what happens - it should resync the array.
Offline
#4 2009-06-16 13:59:29
- oberkeim
- Member
- Registered: 2009-02-23
- Posts: 5
Re: Raid-1 problem
Ok, thanks.
So you also do not trust the DNS323.
Offline
#5 2009-06-16 15:49:13
- fordem
- Member
- Registered: 2007-01-26
- Posts: 1938
Re: Raid-1 problem
It's not a matter of trust - it's a matter of taking responsibility for my data - it's my data and if anything happens to it, I'm the one who suffers.
I replaced a failed disk in an IBM server with 4xRAID1 arrays last week, and what I did there was backup the entire server, remove and replace the failed drive and leave it to resync - exactly what I would do with a DNS-323.
The basic procedures are the same regardless of it's a USD$5000 server or a USD$500 consumer NAS - and - for what it's worth, if you had a drive fail and your data remained available, then your RAID1 array has done it's job - the rebuild/resync task is of no relevance, all that is important is that your data remained available.
Last edited by fordem (2009-06-16 15:50:23)
Offline
#6 2009-06-16 16:00:39
- skydreamer
- Member
- From: At the Atlantic Coast
- Registered: 2007-01-06
- Posts: 232
Re: Raid-1 problem
Just expanding on Fordem's point- one of the most critical issues with any RAID array is the process of rebuilding it after a drive failure.
Imagine that there is a manufacturing problem with the drive batch, which causes them to fail.
Hence as the drive array is being rebuilt the chances are that you may loose the entire array if another disk (the only other one in case of the DNS-323) fails.
There is some interesting reading on RAID5 vs RAID6 reliability- the latter is being pushed by likes of HP, which also applies to RAID1.
(RAID 6 tolerates second drive failure during the rebuilt process.)
Offline
#7 2009-06-16 20:46:18
- mig
- Member
- From: Seattle, WA
- Registered: 2006-12-21
- Posts: 532
Re: Raid-1 problem
fordem wrote:
all that is important is that your data remained available.
IMHO, this is the real value of RAID1 (and often understated in RAID level discussions on this forum).
After a single disk failure, your data remains available such that you can make a current backup,
right before you attempt the recovery procedure of a failed disk.
With all other disk configurations on the DNS-323 you must rely on a periodic scheduled backup, which could result in data loss,
depending on how frequent your backup schedule runs and how frequent your data changes.
DNS-323 • 2x Seagate Barracuda ES 7200.10 ST3250620NS 250GB SATAII (3.0Gb/s) 7200RPM 16MB • RAID1 • FW1.03 • ext2
Fonz's v0.3 fun_plug http://www.inreto.de/dns323/fun-plug
Offline
#8 2009-06-16 22:56:23
- fordem
- Member
- Registered: 2007-01-26
- Posts: 1938
Re: Raid-1 problem
I couldn't agree with you more mig - too many people get hung up on the issues of the how and why of the rebuild, which is not a requirement for redundancy - it's desirable, but not essential.
skydreamer - I think HP prefers the term ADG as in RAID ADG - advanced data guarding and RAID6 is a more generic industry term.
It is not so much the reliability of the one over the other, but the fact that newer technology allows an array with as many as 240 drives, compared to the older limit of perhaps 45 - statistically the chances of a drive failure occurring increases in direct proportion to the number of drives employed.
As drive capacities have increased, the rebuild times have also increased and so in an array with a large numbers of drives, with the increased probability of a drive failure, the possibility of a second drive failing before the first has finished rebuilding, resulting either in loss of data or loss of availability, has become very real - this is the scenario that RAID6 attempts to address.
Whilst RAID6 can theoretically be used with as few as 4 drives, I would say it's not worth the cost until you have maybe a dozen drives or so, and needless to say - discussing reliability of RAID5 vs RAID6 has no impact on RAID1, the ramifications are vastly different.
Last edited by fordem (2009-06-16 23:06:45)
Offline
#9 2009-06-17 14:12:31
- skydreamer
- Member
- From: At the Atlantic Coast
- Registered: 2007-01-06
- Posts: 232
Re: Raid-1 problem
My point was that the RAID1 may suffer a second disk failure during the rebuild therefore taking an immediate backup seeks to avert this scenario.
This is providing that the DNS-323 in question is not used as a backup on its own or being backed up regularly- common sense prevails here.
It is beyond the point of this thread but RAID6 was not introduced exclusively to deal with massive arrays, it is to deal with second drive failure during RAID rebuilt thus increasing server's availability. From time to time I get statistics compiled on drive returns by manufacturer/model/batch from a large sample (100,000+ disks) and for desktop graded hard drives there is is usually a pattern of failures derived from manufacturing difficulties hence the higher chance of multiple drive failure in RAID1 populated with the same disks.
Offline