I have been using 4.7 for a long time and I am ready to buy a license and upgrade to some larger drives but to do this I need a stable version 5 release.

 

Do you guys think we can expect a stable version 5 release in 2013? 

Even though they are called RC I feel 5.0 has been stable for quite some time now.

 

Kryspy

At the rate we're going now, never.

 

I agree though that for 99.999% of people 5.0 RC12a is more than stable in a production system. I've been using it since the early betas without issue.

5.0 has been stable for many months, why the perception that is hasn't ?

5.0 has been stable for many months, why the perception that is hasn't ?

 

Oh I don't know, ya know, maybe the whole "BETA" and "RELEASE CANDIDATE" thing.

Two thoughts ...

 

(1)  Unless you need > 2TB drives, v4.7 work just fine, and I can't really think of any reason to upgrade.    I've got two UnRAID systems; and my original one has been VERY stable since built -- and has been running v4.7 since it's release.    Has only been booted twice in the past year ... after power failures that resulted in automatic shutdowns via the APC UPS package.

 

(2)  v5 RC12a is also very stable.    I've been running v5 for about 5 months, since I built my 2nd UnRAID server using 3TB WD Reds.    Started with ~ RC8 (don't recall for sure what the release # was initially); and have updated every couple of release candidates .. upgrading to v12a after it had been out for a couple weeks.    This system is also VERY stable -- absolutely zero "hiccups" and excellent parity check speeds.    I've run a LOT of parity checks on this system ... one/day for the first week;  weekly afterwards for a few weeks; then again every time I upgraded to the next RC, and 2-3 daily ones after each upgrade ==> and EVERY parity check has been perfect.

 

Note:  Both of my systems are pretty much "plain, vanilla" setups.  Nothing but UnRAID with UnMenu and the CleanPowerDown and APC UPS packages ... nothing else.    The stability of v5 with some of the add-ins may be a slightly different story.    But if all you're looking for is a solid, fault-tolerant storage server, I think v5 RC12a fits the bill -- I suspect it's very likely that this will become v5.0 final  [Just as the last RC in the v4.7 series became v4.7 final].

 

I've run a LOT of parity checks on this system ... one/day for the first week;  weekly afterwards for a few weeks; then again every time I upgraded to the next RC, and 2-3 daily ones after each upgrade ==> and EVERY parity check has been perfect.

 

I honestly wouldn't say that's a good thing, you're wearing the drives down for nothing, I understand the single parity check after an upgrade, and a two to four weekly parity check, but, anything daily is excessive.

 

However, it's up to you.

I honestly wouldn't say that's a good thing, you're wearing the drives down for nothing...

 

as unRAID'ers we have a bit of OCD when it comes to drive usage and the wearing out of drives.

 

As I posted in another thread, I had a number of drives removed from service. A couple were spinning for 8 years without failure.

They were busily used on a daily basis in RAID1 for my linux workstation. These were Hitachi 7200 RPM 9GB SCSI drives.

 

On another machine I had a pair of 10,000 RPM Maxtor SAS drives. (I guess that shows a bit of the Age).

These were spinning from 2005 up until hurricane Sandy.  These were also used in a linux workstation. One partition had a couple of vmware instances, one of which was my XP virtual machine that I never turned off.  History is, when my windows machine died, I loathed re-installing so much, I virtualized XP and used that for the next 8 years under VMware Workstation.  I mention it to give an idea of the usage pattern. A Linux raid1 setup, with some space set aside for VMware with a couple virtual machines running 24x7x365. 

 

Each of these spun 24x7x365 in a fairly medium use environment in interactive workstations and for file serving.  They were not in an air conditioned room.  However I did work on insuring there was adequate air flow.

 

Point is , I would go as far as suggesting that a parity check is not all that hard on drives.

 

The drives are read sequentially from starting track to ending track in a gentle manner.  sector by sector/track by track. No head thrashing. Just reads and/or writes in a sequential manner.  It actually couldn't be any more gentle unless you are reading from/writing to the array while the parity check is active.

Even normal filesystem writes require searching for blocks, journaling, writing data, then updating used blocks, etc, etc.  This causes the heads to move back and forth to each respective location.

 

Reading the drive sequentially from start to finish is about as gentle as can be as long as you have a good power supply and heat is under control.

 

Given that,  I wouldn't say the drives are being worn out.

 

However daily parity checks is probably more then is needed after a person has reached their comfort zone.

What's probably important with all of these checks is consistent speed and sharing that data for unRAID 5's release candidacy.

I agree daily checks are completely unnecessary => I was doing it more to VERY thoroughly test my WD Reds than for any other reason.    I normally just do quarterly checks (not even monthly) -- my new system has had FAR more parity checks run on it than my 4 year old main UnRAID server has 

 

However, I completely agree that a parity check is NOT any significant stress on the drives -- virtually no seeks (just one-track at-a-time), and a very gentle read pattern as it reads from start to finish.    You could do a parity check every day and I doubt the failure rate of the drives would be any different than if you never did one 

I honestly wouldn't say that's a good thing, you're wearing the drives down for nothing...

 

as unRAID'ers we have a bit of OCD when it comes to drive usage and the wearing out of drives.

 

As I posted in another thread, I had a number of drives removed from service. A couple were spinning for 8 years without failure.

They were busily used on a daily basis in RAID1 for my linux workstation. These were Hitachi 7200 RPM 9GB SCSI drives.

 

On another machine I had a pair of 10,000 RPM Maxtor SAS drives. (I guess that shows a bit of the Age).

These were spinning from 2005 up until hurricane Sandy.  These were also used in a linux workstation. One partition had a couple of vmware instances, one of which was my XP virtual machine that I never turned off.  History is, when my windows machine died, I loathed re-installing so much, I virtualized XP and used that for the next 8 years under VMware Workstation.  I mention it to give an idea of the usage pattern. A Linux raid1 setup, with some space set aside for VMware with a couple virtual machines running 24x7x365. 

 

Each of these spun 24x7x365 in a fairly medium use environment in interactive workstations and for file serving.  They were not in an air conditioned room.  However I did work on insuring there was adequate air flow.

 

Point is , I would go as far as suggesting that a parity check is not all that hard on drives.

 

The drives are read sequentially from starting track to ending track in a gentle manner.  sector by sector/track by track. No head thrashing. Just reads and/or writes in a sequential manner.  It actually couldn't be any more gentle unless you are reading from/writing to the array while the parity check is active.

Even normal filesystem writes require searching for blocks, journaling, writing data, then updating used blocks, etc, etc.  This causes the heads to move back and forth to each respective location.

 

Reading the drive sequentially from start to finish is about as gentle as can be as long as you have a good power supply and heat is under control.

 

Given that,  I wouldn't say the drives are being worn out.

 

However daily parity checks is probably more then is needed after a person has reached their comfort zone.

What's probably important with all of these checks is consistent speed and sharing that data for unRAID 5's release candidacy.

 

I'd like to state I never said that it was a drastic wear-down, however, all the less, it is a wear down, even more so, holllyyy crappp doess everythingg takee forreverrr toooo loaddd whileee doinggg aaa parityyy checkkk (Yes, extra letters were required), I get around 1Mbit/s read/write speed during a parity check, I'd hate it if 43.75% of the time I tried to read/write it was slow as turtles (Best simile I could think of).

A parity check is virtually NO wear on the system UNLESS you do other things during the parity check.    Obviously THAT causes excessive thrashing on the disk(s) you're accessing.    When I do a parity check, that's ALL I do => NO other array accesses.

 

  I would say that;

 

VERY frequent parity checks = Safer Data - Hidden data recovery (SMART LEVEL) - fewer parity check errors reported, if any.

 

Rare or NO parity checks = At risk Data - Not enough use for SMART LEVEL repair - Highest amount of parity errors likely reported.

 

  I feel that the best setting would be somewhere in-between the two.  I feel the suggested 30-day interval is good, it is more frequent than should be needed, but spread out enough that a failing drive should actually be noticed, instead of fixed frequently on the fly...

 

  Yes, it comes down to comfort level...

 

  So, based on what your definition of drive failure is, a more frequent, (daily) parity check could have less data loss, and seem to be a good drive... while one that is not checked as often may get replaced sooner, even though it may actually be more healthy than the drive still in use in the daily parity check scenario.

 

  I have seen with some drive faults, that running a parity check too frequently, will repair (heal/relocate data) without the use or unRAID ever knowing about it.  (but that was the whole idea anyway...)

 

  So no data lost, and with some drive firmware the re-allocated sector count information does not show either!  Some drives do NOT start to even increment/decrement allocation counters untill the number of spare sectors/clusters reaches some pre-determined level... typically in such instances the number that is equal or near the counter value.  Many newer drives have FAR more spare space available than would be expected from looking at the SMART values.  So untill the drive degrades to those numbers, (which from most manufactures were set as vendor default counts years ago before we had LARGE drives), the numbers will not change.  If you look at SMART values frequently, you may see a pending sector re-allocation event, then later see it is back to ZERO, with no change in the allocated event table...

 

  If you wait longer between the parity checks, it is more likely that you may see the drive faults as they increase in severity, so you can tell that the drive actually is in need of some attention, even if it is normal SMART self-heeling...  The point is that at this point you may want to increase the frequency again of a parity check, to keep the data integrity higher, and allow SMART to handle things at the drive level.

 

  If there is a true reliability issue with a drive, it would be better to find out sooner, instead of later.  That way the drive may still be under warranty and more data may not yet be on the drive...  Until I started using unRAID, ALL my hard drives were always set to ALWAYS spin, I never would shut down a drive if the computer was on, and I would seldom shut down a computer either.  I have seen more cases of drive failures not coming back after a power down than anything else, (and fan failures for that matter also) due to bearing failure, but that has been after MANY YEARS of non-stop use.  So it is all a matter of compromises, whaich type of failure can you tolerate, and where the best average compromise may be for a given application.

 

... I have seen more cases of drive failures not coming back after a power down than anything else, (and fan failures for that matter also) due to bearing failure, but that has been after MANY YEARS of non-stop use. ...

 

Agree.  I've also been doing this for MANY years (my first hard drive was a Seagate 26MB 14" Winchester -- circa 1980 for the bargain price of $4500).    The most common causes of failures in modern drives are power fluctuations (which you can protect from with a good UPS with AVR); and excessive power cycles.  The rated power cycle count has gone up a good bit in recent years, so spinning down your array is fine with any reasonable time setting (I use 2 hours, but 1 hr is fine).  Drive are typically rated for 10,000 or more power cycles.

 

I have seen more cases of drive failures not coming back after a power down than anything else, (and fan failures for that matter also) due to bearing failure, but that has been after MANY YEARS of non-stop use.

 

Indeed - I used to be busy for a couple of days after the Christmas/New Year break.  Everything was turned off for this long break after having been running for the previous 350 days, and a number of hard drives would no longer spin up.  If the end of the platter shaft was exposed, I could sometimes get the drive spinning again.