For those using NFS...

[Russ Uno]

What software are you using that won't run under 10.9?

 

What I said was it would be too costly to upgrade just to run 10.9.

On one Mac it would cost me about $2400 to upgrade to 10.9 compatible CS graphics and Data Base software.

On another another Mac I'm running ProTools 10 HD TDM that just cost me $3000 in December to upgrade and it is not supported by Avid on 10.9 yet and may never be since they are moving to ProTools 11 which is an $8500 upgrade that obsoletes my current ProTools hardware, about $12000 worth.

 

Plus weeks of pain getting everything stable again.

When a OS is upgraded and key software upgrade there are any number of secondary software that must be upgraded as well, $$$$.

[NAS]

Apple are in the process of moving away from AFP so only old Apple products will need it. It stands to reason that if it benefits unRAID to drop it in the v6 series we should.

 

Users who are stuck using AFP would stick to v5.

 

That is the nature of the beast when using less mainstream protocols unfortunately

[ironicbadger]

Apple are in the process of moving away from AFP so only old Apple products will need it. It stands to reason that if it benefits unRAID to drop it in the v6 series we should.

 

Users who are stuck using AFP would stick to v5.

 

That is the nature of the beast when using less mainstream protocols unfortunately

 

We should only drop it once Apple have. Leaving people without a choice 'hoping' Apple do is not a smart move.

[madburg]

Lets leave it until that separate post turns up then. I dont care about AFP and i am sure lots of NFS people like me have no interest in reading about AFP in the NFSv4 thread

For someone who has no interest in reading about it (nor use's AFP), you have alot to say.

 

Whats the benefit to dropping it? There is no benefits to dropping AFP at this time, Apple isnt going to get everything working overnight and neither are the apps we work with. Like many of you like to say, if your not using it then dont, and dont enable it, how does it bother you.

 

I've been waiting for 64 bit unRAID and the lastest AFP bundle, which Tom stated AFP would be upgrading in v6, so no i would be very unhappy with having to stay on v5 and older AFP. Easy for someone to state when they have no care or use for it.

 

If Tom complied and added the updated AFP in the next v6 Beta, his promise is kept and things would move on, instead of these ramblings.

 

[NAS]

Who cares about AFP its a dieing protocol not even endorsed by the people who made it anymore.

 

Thats the way of the Apple.

 

Just use V5 its continuing to be developed and there is no EOL for it and will likely outlive Apple support.

[madburg]

Who cares about AFP its a dieing protocol not even endorsed by the people who made it anymore.

Who cares? the people requesting it be update in unRAID and have dependancies on it.

Endorsed? What are you talking about, it is supported today by them and will be until the day they finish with the dependancies and third party dev make the move as well. That isnt happening today nor tomorrow.

 

 

Just use V5 its continuing to be developed and there is no EOL for it and will likely outlive Apple support.

But Tom clearly stated he would not compile the lastest version and add it in the v5 series and that he would do it with the start of v6. So once again people make crap us as it pleases them. Im going by what the developer of the product stated, which just threw a curve ball by entertaining the notion to back pedal on AFP altogether.

 

Drop all versions of NFS in v6, I dont use it and see no benefit, what you want it? use v5 then.

After all Tom couldn't get NFS3 working properly, like other companys have. Wasting time to see if he can get NFS4 to work, I see no benefit.

Just this sound better and fair to you?

 

 

 

[NAS]

Things change. The world moves on.

 

Now... all arguments that V5 should continue to support AFP for the life of the protocol or the product are absolutely valid. That was a feature added to unRAID and invariably some people would have purchased unRAID either solely for this reason or at least in passing.

 

I have absolutely no issues with that and strongly back this argument.

 

However I see no reason why the new product, namely V6, must have AFP in light of what is happening with the protocol in general. It is a dieing protocol and if it is added to V6 you will never get rid of it. Someone will always complaining that their expensive Apple product needs it so they must have it.

 

I would be pissed if i was in your boat. I would be arguing tooth and nail that v6 needed AFP because i would want v6 goodness and the ability to run my old kit the way I always have....

 

... but I still say that v6 is no place for AFP.

 

[madburg]

Your at liberty to say what you want, I stated my piece as well. Just remember since your not using it you shouldnt be so quick to express your view on those who do.

 

[NAS]

sure i should that the whole point of this forum.

 

But let me jump to the other side of the fence for a bit and come up with a real world product

 

 

What product uses AFP, cannot run, or runs terribly with NFS or CIFS and needs a 64 bit unRAID kernel?

 

 

[madburg]

Sure no problem, keep blabing.

 

Not what product, its unRAID that takes a crap with products running under AFP, if parity check is running you cant access anything at all.

Tom knows all this, we had discussions, logs submits. I couldnt care less what you think nor owe an explanation or prove anything to you.

 

What product use NFS3 and cannot run that NFS3 needs to be dropped?

[prostuff1]

I would like to see NFS stick around for now.  Perhaps in version 6.1/6.2/6.5/7.0 AFP can and should be dropped.  I think it is to early to do that now.  That is coming from a person that uses OS X almost entirely at home but still use SMB for most all the connections.

 

until Apple moves Time Machine away from needing AFP it should stick around.

[NAS]

Sure no problem, keep blabing.

 

Not what product, its unRAID that takes a crap with products running under AFP, if parity check is running you cant access anything at all.

Tom knows all this, we had discussions, logs submits. I couldnt care less what you think nor owe an explanation or prove anything to you.

 

What product use NFS3 and cannot run that NFS3 needs to be dropped?

 

I think you lost the debate right there. Well done

 

I would like to see NFS stick around for now.  Perhaps in version 6.1/6.2/6.5/7.0 AFP can and should be dropped.  I think it is to early to do that now.  That is coming from a person that uses OS X almost entirely at home but still use SMB for most all the connections.

 

The userbase of NFS will be high and both v3 and v4 are no where near EOL so we need to be really careful with this one. I would like to hear real world exmaples of devcies that cant support v4. I suspect most will be cheap firmware based media players but I could be proven wrong.

 

until Apple moves Time Machine away from needing AFP it should stick around.

 

Agreed. I would argue that ensuring LTS support on v5 fulfills this requirement.

 

 

Edit: The debate  got quite passionate and heated there with nor real technical content. Putting this thread under the heavy moderation announcement rules. Fork if you want to ramble.

[madburg]

You start talking crap about AFP in this thread, when you yourself stated to wait from Tom on a AFP thread, now you are no longer happy so your deleting post and now state "The debate  got quite passionate and heated there with nor real technical content"? Really?

[NAS]

You start talking crap about AFP in this thread, when you yourself stated to wait from Tom on a AFP thread, now you are no longer happy so your deleting post and now state "The debate  got quite passionate and heated there with nor real technical content"? Really?

 

No more OT here please fork to a new thread to ramble.

[butlerpeter]

Could it not be possible to remove AFP from the main release but make it available as a plugin?

 

Surely that's exactly the type of thing plugins are designed for - making additional functionality available so that users who want it can install it.

[NAS]

I suspect AFP will have to go one of the following ways:

 

Stay as is (IMHO the most likely due to the backlash but wouldn't be my personal design choice)

Stay included but not supported. This would likely mean no native emHTTP configuration. I think this may be the best compromise.

Not included at all and available via an addon. Not even sure this is completely possible but if it was would let the community who need it own it.

Present AFP via a VM sitting on top of unRAID. This is my personal favorite as it would allow unRAID to focus on the core deliverables and AFP advocates to use a VM distro that actively supports AFP.

 

 

 

[madburg]

Stay included but not supported. This would likely mean no native emHTTP configuration. I think this may be the best compromise.

Not supported? It not as simple as enable or not, you have to have the means to set AFP Security setting per disk and/or TM settings as well, with no native emHTTP support, how so? Only thing some of us do and Tom didn't provide is set the DB to a non array disk so we don't lose it and its data when a reboot is require of unRAID.

 

Not included at all and available via an addon. Not even sure this is completely possible but if it was would let the community who need it own it.

No interest to use anything someone outside of Tom attempts to make. IF so same should be done with NFS, make it an addon.

 

Present AFP via a VM sitting on top of unRAID. This is my personal favorite as it would allow unRAID to focus on the core deliverables and AFP advocates to use a VM distro that actively supports AFP.

No interest in Xen nor VM, need AFP at the unRAID layer as all supported protocol have been. IF so, off load NFS to a VM as well.

 

[Neo_x]

guys...... /girls?

 

may i recommend that we stay on topic...hence the title :

For those using NFS..

 

AFP was shortly touched on, and i am very sure Tom sill start a new thread

For those using AFP.. , when time is right ( his hands will be more than full enough with all the current work, and thus reading through a whole AFP discussion 3 pages long in an NFS thread will just be a to much.

 

 

reading through the history of the NFS problems, and points listed in this thread, its a no-brainer. Move to NFS V4 only with the new unraid. If he keeps it on disk shares, the users still have a work around if they must use it ( most media players will still build a single movie database even if you give it disk1/movies , disk2/movies etc etc.)

 

the ESXI part though is a bit troubling (if somebody has a working ESXI system using unraid...). maybe get a poll going to get a better range of feedback (ESXI can also point to the disc share with the data store on, although more cumbersome).

 

keep up the good work tom!

 

 

[NAS]

I would suggest the best way to see the impact of dropping NFSv3 is to just do it in the next beta.

 

A beta test cycle is the perfect place to gauge the real world impact rather than hope people read this thread before the ship has sailed. break it and they will come.

 

As for AFP this thread is now both. I would have renamed it already if it wasnt Toms, I am sure he will soon..although there is little debate on the subject it is either "must have because always have" or not.

[madburg]

I would suggest the best way to see the impact of dropping NFSv3 is to just do it in the next beta.

 

A beta test cycle is the perfect place to gauge the real world impact rather than hope people read this thread before the ship has sailed. break it and they will come.

 

As for AFP this thread is now both. I would have renamed it already if it wasnt Toms, I am sure he will soon..although there is little debate on the subject it is either "must have because always have" or not.

 

Problem maybe that not enough people have the equipment and/or time to test a beta like this at this point because its not recommended or supported on a production system. So as an example someone in a country who spent a great amount of money (thats how pricing goes in some countries) on a media player with NFS3 support only and only has 1 system can't chime in (not in the know). Or someone with a complex system with say one component being attaching an esx server(s) to unraid but doesn't have the time to follow the threads only realizes later as v6 comes out of beta (just an example here) what changed (removal of NFS3). Not a happy camper.

 

This is why i firmly believe a v5 counterpart as the first v6 (64 bit) would go a long way to keep the vast client base happy. I can't speak for Tom, but if he was willing to work with 3 forks (made up labels here->) v5-32bit/v6-64bit/v6.1-64bit-Xen/Samba4/NFS4/No AFP/NFS3 whatever else, would be great. Or without knowing why he doesn't wish to have NFS3&4/AFP keep going, so for now with v6 keep them going would be another option. I (just my opinion) don't believe never giving what was already in v5 in a 64 bit flavor is the way to go, and tell people stay on v5 if thats what you want, not that he official made that stance, but the remarks being made is so he is aware to hopefully make the right call.

 

Many of us are still trying to put something together hardware wise to even begin testing, i'll be starting v6 in a VM soon as the first round of testing, as my HW and budget are coming together at a slow pace personally.

 

[trurl]

v6beta3 has unRAID without Xen in the boot menu. Do we think that will be going away?

[PeterB]

Would it be possible to create the release with both both NFS v3 and v4 available, but with v3 disabled by default?

 

Those who can work with v4 will do.  Those who need v3 can enable it (by a setting in extra.cfg?), with warnings that they will be exposed to 'stale file handle' errors.

[boof]

I would be cautious about removing nfsv3 support :

 

- Some clients don't support it. Cheap media players for sure and even some bigger names like, as has been mentioned, esxi.

 

- Though it's 'just' a single release increment it's actually quite a drastic change in architecture and behaviour, it's not a small or incremental leap it's borderline completely different. hence the above compat issues.

 

- There is nothing really wrong with v3. The issues are how the user share system works. This is unique to unraid afaik. The rest of the world trucks on with nfsv3 without too many issues - the main argument for v4 is usually better security (particularly when used with kerberos integration). I have no issue with 'unraid would work better with nfsv4' but blaming it on shortcomings in nfsv3 is a bit iffy. It's the fact unraid changes the id's under the hood that causes problems.  I manage a multi PB NAS system at work which exports via nfs3. It's a multi frontended system with a multi-tiered parallel filesystem on the backend and nfs behaves splendidly. Saying in the modern world of tiered fs's etc NFS v3 can't cope isn't correct - it's all down to the implementation at the filesystem.  I suspect this is an argument / can of worms that is completely pointless having / opening though.

 

- At the very least flip the default behaviour to v4 but allow enablement of v3 with grave messages of 'at own risk'. Removing support entirely (for user shares) seems a heavy handed approach. Or fix shfs

[NAS]

I agree we should be cautious but...

 

I would be cautious about removing nfsv3 support :

 

- Some clients don't support it. Cheap media players for sure ....

....

 

TBH i dont think  this is a good argument. People who purchased firmware based media players should accept they get old the minute they stop releasing firmwares (which in my experience is almost immediately especially with the cheap ones).

 

Old kit stick with v5 which is still a stable and supported product.

 

.... as has been mentioned, esxi.

....

 

IMHO there is no scenario where esxi wont end up with nfsv4 support. So again until that time v5 is viable and then when it does support NFSv4 people wont care anymore.

 

 

I am not trying to be argumentative but v6 is the future and v5 is still completely proven product.

 

If for some reason we cant or shouldnt have both v3 and v4 in in unraid 6 (which obviously is since Limetech started this thread) then the only choice is NFSv4 for unraid v6 and those that cant use it stick with unraid v5

 

[boof]

I agree we should be cautious but...

 

 

TBH i dont think  this is a good argument. People who purchased firmware based media players should accept they get old the minute they stop releasing firmwares (which in my experience is almost immediately especially with the cheap ones).

 

Old kit stick with v5 which is still a stable and supported product.

 

Unsurprisingly I would disagree - many people will make choices based on the feature set of the player. NFSv3 is a robust, reliable,stable and most importantly ubiquitous protocol and it's sensible to have bought something that supports it. How many other NAS appliances / devices / OS's don't support NFSv3? I don't know the answer but my hunch would be unraid would be quite unusual in this.

 

Outdated firmware is neither here nor there - as above v3 is pretty robust so shouldn't need much fettling on a client. It's clear that unraid 5 isn't stable or supported with nfsv3 and user shares though - so what are people to do?

 

IMHO there is no scenario where esxi wont end up with nfsv4 support. So again until that time v5 is viable and then when it does support NFSv4 people wont care anymore.

 

Not roadmapped as far as I know. I'm not sure if it's a chicken and egg thing with storage vendors, regardless it's not there.

 

I am not trying to be argumentative but v6 is the future and v5 is still completely proven product.

 

As above given this is the biggest spate of unraid progressive development in quite some time it seems ironic that a side effect is to drop nfsv3 support on an important feature. That's not a good future particularly when v4 and 3 can co-exist perfectly happily fullstop and certainly no worse than they *do today* v3 wise on unraid. So in fact what this is really an admission of is that unraid doesn't robustly support nfsv3 for user shares at all either in 5? That's not making unraid 5 a proven product and one we should stick with to keep using nfsv3 on! Comes back to one of the most standard protocols on the planet not being supported by one of unraids headline features..that's not progression.

 

Unraid 6 being the future isn't tied to unraid stopping support of nfsv3 with user shares. Surely you'd want the future to be unraid 6 supporting nfs 3&4 across the board?

 

If for some reason we cant or shouldnt have both v3 and v4 in in unraid 6 (which obviously is since Limetech started this thread) then the only choice is NFSv4 for unraid v6 and those that cant use it stick with unraid v5

 

They'll both be there - just not for user shares by the sound of it. Though I have an expectation a quick nip under the hood will fix that anyway with the same caveats for v3 support as current. Probably a day one plugin from the community that this forum will end up supporting. Those that can't use v4 for user shares will officially have to stick with unraid 5 and the now officially acknowledged(?) broken implementation of shfs and v3? That's no solution as it means there is no stable solution. Fix shfs, removing v3 support for user shares is just ducking the issue and removing a chunk of functionality of the product. At a time when the development pace of unraid is almost unheralded (certainly in my time here) chucking out features seems a backwards step.

 

AFP and netatalk I'm less inclined to have an opinion on. It's a different beast than NFS (certainly far more niche! especially outside macs) but I'd be surprised if AFP deprecation wasn't more to do with the fact that there seems little point in apple maintaining their own cifs stack and afp to do the same thing. And cifs versus afp is a no brainer. Anyway it's a different situation - you're facing a protocol (AFP) being formally killed off by it's primary (only?) vendor. Following suit seems sensible. That's not the case for NFSv3. I also have a sneaky hunch that most people only want AFP (via netatalk) support on unraid for time machine support which presumably will be fixed to work well with CIFS if Apple are also going to have to rely on the same (though given apple do their own cifs stack they might diddle it just for that I suppose..)

 

I don't know how speculative this is from limetech though the premise of the this thread was to seek opinions. So there is mine. Lack of nfsv3 support in unraid for one of it's headline features / USP's would be a step backwards in my opinion. And if we are to infer from this that it means unraid 5 also has no stable support for nfsv3 and user shares with no path to fix then that's also troubling.