maybe jumping ship?

[evans036]

i need to start using 3tb drives and generally dont trust beta software and am concerned we may not see v5 go GA for a long time.

 

and before folks start telling me that the beta releases are stable, i cannot ignore the message posted by the author of the software that the beta should not be used for anything but test arrays.

 

so, i am thinking about jumping ship to snapraid. does anyone have experience (ie pros/cons) with it?

 

here's a link: http://snapraid.sourceforge.net/compare.html

 

btw, i use unraid to store static data like movies & backups. i can live without real time parity protection (as long as i get it within 24 hours or so).

 

thanks a lot,

 

steve

 

[prostuff1]

biggest con and the reason I would not use snapraid is because it is not realtime.  I do not want to have to think about having to initiate protection every time I copy data to the machine.

 

 

Also, I run 2 x 5.0b14 production machines and they are working just fine.  One is an HP Proliant N40L machine that has been running 5.0b14 from day one and the other is a ESXi build that has been updated from 5.0b2->5.0b6a->5.0b14.  Never had a problem with any of them honestly.

 

The general concensus is that 5.0b14 and 5.0b12a are pretty stable at this point and the only difference depends on what hardware you have.

[Johnm]

I agree with prostuff.

 

I have been running massive arrays on beta. but you are right, some people are uncomforatable with the term beta. the warnign is there because the first few people to jump on the beta might have things break. the latest betas hae been out so long, we know it's limit. the beta people are considering them "stable" for most of the features. I'm not trying to convince you to hop on the beta, just point out it is so close to ready... Even Tom thought it would be RC in March.

 

Honestly, the biggest reason that 5.0  is not released at this point is out of the hands of Limetech. The issues are kernel issues at this point effecting primarly drivers for certain HBA's and NFS shares.

 

Yes, there might be a few minor software glitches that we are unaware of at this point.

Everytime Tom updates the kernal to add new features thinking we are about ready for a RC, something new pops it head up. Right now it is the LSI driver, NFS and AD being broken.

 

If you are honestly thinking about jumping ship, I would suggest you toss together a test server and try whatever you are thinking about. Nappit, freenas, flexraid, snapraid, ect.

 

and as prostuff pointed out, snapshots can be a little risky if you dont keep up with it and can be quite time consuming (24 hours for each snapshot with the raid5 option).

I would also be scared of having a drive starting to fail then you start the snapshot and the drive dies mid-snap... then what? you have no recovery options?....It looks like you are SOL unless i read that wrong.

 

It looks like snapraid was really geard for a few disks, not 20 drives. Snapraid does get slower and slower the more drives you add. I also noticed that it takes 20ish hours to recover any lost files/folders in snapraid.. EEK. You will also have to have "beast hardware" for snapraid to work quickly since it is doing heavy lifting.

 

I would read the entire (small) support forum and see what issues and problems you have to be aware of. that's where you'll get the best answers.

It looks to be a single person with limited tech support..  it is your call. lets us know how it works if you do try it.

 

I'll admidt, I am also considering a faster method of backing up my main unraid (not replacing it). so far i have not found anything faster thats comparible to unraid. i really want to keep low power use and 100% protection. so far nothing.. so i'll just keep on chugging along. hardware raid is my best option atm and i have to many mismatched drives for that. it would cost me $2k to upgrade Goliath to a 40TB hardware Raid6.

[evans036]

excellent replies.

 

thanks guys.

 

steve

 

[brian89gp]

I wouldn't necessairly just assume that since there is no "beta" tied to the SnapRAID versioning system that it is more reliable then unRAID beta.  Look at the the release notes to the 1.9 release.  Its great that it "improves the chances of recovering data after an aborted 'sync' ", but think about what that implies.  You can loose data with an aborted sync?

 

I have looked at SnapRAID too, for the same type of storage requirements you are referring to.  It is an interesting concept but I think it is still in its beginning stages.  6-12 months from now, who knows.  If the concept delivers then it might be a serious contender for a lot of products.  The hashing for data integrity has my eye for sure.

 

I have been running 12a inside of ESXi, and it has been hard power cycled more then a dozen times (CPU overheating and locking up ESXi).  Each time, unRAID came back with no errors and no data loss.

[opentoe]

i need to start using 3tb drives and generally dont trust beta software and am concerned we may not see v5 go GA for a long time.

 

and before folks start telling me that the beta releases are stable, i cannot ignore the message posted by the author of the software that the beta should not be used for anything but test arrays.

 

so, i am thinking about jumping ship to snapraid. does anyone have experience (ie pros/cons) with it?

 

here's a link: http://snapraid.sourceforge.net/compare.html

 

btw, i use unraid to store static data like movies & backups. i can live without real time parity protection (as long as i get it within 24 hours or so).

 

thanks a lot,

 

steve

 

 

Lookey lookey, same boat as I was in. I just couldn't trust 20TB of data with a beta so I went out and spent money on a Synology box. The DF2411 to be exact. I still have my unraid box, which I use as a backup to my primary Synology box. I can now edit movies with no pauses and freezes...just because of the speed increase. If unraid were ever to come out with a 5.0 final, I would probably still get it...to upgrade my current one.

 

 

[evans036]

yeah - i do get the sense that Limetech are on the one hand saying: 'use the beta because it is really stable' but on the other hand saying 'only use it for test arrays'.

 

definitley mixed messages

 

and in a professional enviroment i could never condone use of beta software to protect data.

 

 

[boof]

5 beta has (in various versions) hardware issues, nfs problems, ad problems and I've seen various config glitches during initial setup (though due to hardware requirements I'm running 12a so later ones may have ironed the unraid specific issues I saw out). Not many of these are specific unraid issues, but kernel problems - doesn't really help when you're stuck with what's bundled and need a stable platform though.

 

Whilst your data is likely safe with 5 (I'm running it myself with no problems data wise), presuming you can match the hardware requirements I do understand the reluctance at jumping on the 5 ship just now. And I agree about the mixed messages especially as there are definite holes in 5 right now.

 

I tested snapraid, not much to add what's already been said though in my (limited granted) testing not much went wrong with it. And, like unraid, at the end of the day you still have your data on 'normal' filesystems. And parity in whatever format should never be a substitute for backups.

 

My main falling point was that snapraid in itself was fine but you would also lose unraids user share 'single name space with data placement rules' technology. It was really that which put me off snapraid. mhddfs came closest but the details of how it works in some circumstances of data placement made it look like it would be less fire and forget than user shares. I couldn't find many other useful alternatives (unionfs / aufs were probably the closest) but that may not be an issue if you don't need user shares or don't mind manually managing your data a bit more.

 

Flexraid is the other obvious candidate and it also has a realtime engine and user shares-esque concept (though last time I tested I could break the linux release inside of 2 seconds). I'm not sure I'd recommend it but it's an option and there are certainly 'interesting things' happening with it right now which could, optimistically, turn out to be better for the project as a whole.

[Fluid]

I left for Synology.  Higher up front cost, but the interface and features rule.

[evans036]

5 beta has (in various versions) hardware issues, nfs problems, ad problems and I've seen various config glitches during initial setup (though due to hardware requirements I'm running 12a so later ones may have ironed the unraid specific issues I saw out). Not many of these are specific unraid issues, but kernel problems - doesn't really help when you're stuck with what's bundled and need a stable platform though.

 

Whilst your data is likely safe with 5 (I'm running it myself with no problems data wise), presuming you can match the hardware requirements I do understand the reluctance at jumping on the 5 ship just now. And I agree about the mixed messages especially as there are definite holes in 5 right now.

 

I tested snapraid, not much to add what's already been said though in my (limited granted) testing not much went wrong with it. And, like unraid, at the end of the day you still have your data on 'normal' filesystems. And parity in whatever format should never be a substitute for backups.

 

My main falling point was that snapraid in itself was fine but you would also lose unraids user share 'single name space with data placement rules' technology. It was really that which put me off snapraid. mhddfs came closest but the details of how it works in some circumstances of data placement made it look like it would be less fire and forget than user shares. I couldn't find many other useful alternatives (unionfs / aufs were probably the closest) but that may not be an issue if you don't need user shares or don't mind manually managing your data a bit more.

 

Flexraid is the other obvious candidate and it also has a realtime engine and user shares-esque concept (though last time I tested I could break the linux release inside of 2 seconds). I'm not sure I'd recommend it but it's an option and there are certainly 'interesting things' happening with it right now which could, optimistically, turn out to be better for the project as a whole.

 

nice reponse.

 

the 'unified' share name is a feature that really appeals to me and is frequently overlooked when comparing features with other products like snapraid.

 

prior to unraid, i would spend a lot of time trying to manage multiple shares across mutliple hosts for my backups & movie collection. that was a pain.

 

life is a lot simpler for me now that i have unraid and two shares (one for backup & the other for movies).

 

btw, i did finally upgrade to version 5 with no issues.

 

steve

 

[evans036]

I left for Synology.  Higher up front cost, but the interface and features rule.

 

thanks for the info. i'll check it out.

 

steve

 

[mbryanr]

Another option is to use a smaller Syno in conjunction with unRAID.  Get the same UI benefits with a much lower upfront cost. 

[Dougy]

I left for Synology.  Higher up front cost, but the interface and features rule.

 

i see a 12bay Synology for around $1400aud.  does it allow a mix of disk size and provide parity with just one disk like unraid?? 

[mbryanr]

normal RAID configurations, some allow for "SHR" -Hybrid Raid with a mixture of drive sizes.

http://www.synology.com/us/products/features/RAID.php