LinuxQuestions.org
Download your favorite Linux distribution at LQ ISO.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Networking
User Name
Password
Linux - Networking This forum is for any issue related to networks or networking.
Routing, network cards, OSI, etc. Anything is fair game.

Notices


Reply
  Search this Thread
Old 12-20-2015, 11:56 AM   #1
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Rep: Reputation: 0
Unhappy NFS inefficient link


I have been trying to link two servers together over NFS.

But when I try to download a file (to my computer) located on one server through the NFS shared mount on the second server, the bandwidth required is about 3x that of the regular bandwidth that would be required through a direct e.g. scp link.

Is this normal?

E.g., I mean, my server that shares the files has a 500KB/s upload. But when I retrieve that file through the other server (that provides a cloud service) what happens is this:

- total transfer throughput is about 250KB/s
- origin server uses 500KB/s up, but there is only 250KB/s real transfer
- NFS overhead for this 250KB/s upload seems to be 250KB up and 250KB down.

In the end a 250KB/s upload consumes 500KB/s up and 250KB/s down from the perspective of the origin server.

This mean the bandwidth required is effetively tripled.

Is this normal?

Is this expected?

Is there anything I can do about this? This makes it pretty nonsensical to share remote files over NFS. The link between origin server and exposure server happens through OpenVPN, but that shouldn't matter much in terms of throughput.

It is pretty bad. I do not know of any other protocols (except CIFS/SAMBA) that would apply. I haven't tested CIFS/SAMBA yet. I thought NFS would be the obvious way to go between Linux systems but this is pretty poor.
 
Old 12-20-2015, 12:13 PM   #2
TB0ne
LQ Guru
 
Registered: Jul 2003
Location: Birmingham, Alabama
Distribution: SuSE, RedHat, Slack,CentOS
Posts: 26,608

Rep: Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960
Quote:
Originally Posted by xenmaster View Post
I have been trying to link two servers together over NFS.
But when I try to download a file (to my computer) located on one server through the NFS shared mount on the second server, the bandwidth required is about 3x that of the regular bandwidth that would be required through a direct e.g. scp link. Is this normal?

E.g., I mean, my server that shares the files has a 500KB/s upload. But when I retrieve that file through the other server (that provides a cloud service) what happens is this:

- total transfer throughput is about 250KB/s
- origin server uses 500KB/s up, but there is only 250KB/s real transfer
- NFS overhead for this 250KB/s upload seems to be 250KB up and 250KB down.

In the end a 250KB/s upload consumes 500KB/s up and 250KB/s down from the perspective of the origin server. This mean the bandwidth required is effetively tripled. Is this normal? Is this expected?

Is there anything I can do about this? This makes it pretty nonsensical to share remote files over NFS. The link between origin server and exposure server happens through OpenVPN, but that shouldn't matter much in terms of throughput.

It is pretty bad. I do not know of any other protocols (except CIFS/SAMBA) that would apply. I haven't tested CIFS/SAMBA yet. I thought NFS would be the obvious way to go between Linux systems but this is pretty poor.
NFS is (typically) slower..maybe...sometimes...depending on the setup. There are FAR too many variables to definitively say yes or no. The version of NFS is one thing...the latest versions are typically a bit faster, and if you're going from NFS 2 to NFS 4, you may see issues. But other things are in play, such as write latency. See if mounting the NFS share in async mode helps any.

Using rsync over SSH, or just SCP, the remote process writes asynchronously...but if you have NFS mounted in sync mode, the writes aren't confirmed immediately, and the NFS server keep shoveling data in, until it hits disk/controller cache limit, before confirming to the NFS client that the write was successful.

If async helps...there is something to take into consideration. If something happens to the copy mid-stream, you will most probably end up with inconsistent data on disk.

Personally....I avoid NFS like the plague, if I can avoid it. I can mount remote systems with SSHFS, and greatly simplify the firewall rules, services running, etc. Unless you're beating up the share, the CPU load from that type of mount should be acceptable. If you're only looking for backups/security/some-type-of-safety, then setting up an rsync over ssh job nightly may work too.
 
1 members found this post helpful.
Old 12-20-2015, 12:33 PM   #3
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Original Poster
Rep: Reputation: 0
Thanks for the fast reply.

I was writing some more info about the systems:

The server is a Synology NAS. I assume it is using UDP but I'm not sure. The client is a Debian 8 machine. These options around on the server:

udp_read_size=8192
udp_write_size=8192
nfsv4_enable=no

So I assume the server is using version 3.

These are my mount options on the client:

(rw,relatime,vers=3,rsize=131072,wsize=131072,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys, mountvers=3,mountport=892,mountproto=udp,local_lock=none)

excluding the address. These are defaults.

It seems the client wants to use as much larger rsize/wsize than the server specifies.

The server that exposes the service (cloud service) has limited disk space so I am using a slower server but with larger disk space as a "source". Currently it has a slow link (slow uplink) but even if I were to use a faster disk server, I would still be confronted with the same problem, mostly.

But if async makes everything unreliable, that is not really good. I can't test it yet, doing some transfer.

So basically SSHFS is a way to mount a remote filesystem that is exposed through SFTP on the server. I'm not sure if I want that. NFS is not great but it seems to be fine and the overhead is just too large.

Maybe I can increase the buffer sizes on the server? I see there are some ways to test throughput.
 
Old 12-20-2015, 02:02 PM   #4
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Original Poster
Rep: Reputation: 0
I realize I may not be able to reproduce the low transfer speeds I had earlier.

But I'm not sure yet. In my latest test I had the maximum download speed (to my computer) I could theoretically get, more or about. It was the full '500' KB/s. Much different than the 250 to 300 I had before.

Still looking to increase buffer sizes but it mostly uses TCP for transfer. Thank you for your help though, already.
 
Old 12-20-2015, 02:41 PM   #5
TB0ne
LQ Guru
 
Registered: Jul 2003
Location: Birmingham, Alabama
Distribution: SuSE, RedHat, Slack,CentOS
Posts: 26,608

Rep: Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960
Quote:
Originally Posted by xenmaster View Post
Thanks for the fast reply.
I was writing some more info about the systems: The server is a Synology NAS. I assume it is using UDP but I'm not sure. The client is a Debian 8 machine. These options around on the server:

udp_read_size=8192
udp_write_size=8192
nfsv4_enable=no

So I assume the server is using version 3. These are my mount options on the client:

(rw,relatime,vers=3,rsize=131072,wsize=131072,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys, mountvers=3,mountport=892,mountproto=udp,local_lock=none)

excluding the address. These are defaults. It seems the client wants to use as much larger rsize/wsize than the server specifies.

The server that exposes the service (cloud service) has limited disk space so I am using a slower server but with larger disk space as a "source". Currently it has a slow link (slow uplink) but even if I were to use a faster disk server, I would still be confronted with the same problem, mostly. But if async makes everything unreliable, that is not really good. I can't test it yet, doing some transfer.
You are correct, which harkens back to, is the CLIENT using NFS3 as well? And as said, async doesn't make it unreliable (and I never said that). What it *DOES* do is make your writes faster...the only way it will corrupt a file, is if the link is broken mid-stream, or the server/client goes down during a transfer. So, exactly the same thing as would happen over SSHFS, FTP, or pretty much any other transfer method.
Quote:
So basically SSHFS is a way to mount a remote filesystem that is exposed through SFTP on the server. I'm not sure if I want that. NFS is not great but it seems to be fine and the overhead is just too large.
SSHFS isn't bad at all. As said, if you're going to be transferring a LOT of data, all day, every day, it may become an issue. If you're doing sporadic reading/writing, it's not going to make much of a difference at all.
Quote:
Maybe I can increase the buffer sizes on the server? I see there are some ways to test throughput.
Perhaps, but you may be limited on what the NAS will let you do from that standpoint.
 
Old 12-20-2015, 06:58 PM   #6
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Original Poster
Rep: Reputation: 0
Well in principle if I do any filetransfer that I know might get broken, I use rsync. Although usually HTTP downloads continue well, unless the browser that is doing the downloading deletes your files upon connection break-u p ;-). NFS is like this persistent filesystem that I can see from the exposure server. It works reasonably well although I didn't do much with it since that episode with the low transfer speeds.

It's nice to be in a cloud interface and browse your files that way.

I forgot; the VPN used is over TCP which is why the traffic showed as TCP traffic even though it uses UDP. I would need a second VPN server instance to allow UDP. Not that I've ever seen it lower my transfer rates.

Still don't make me feel well about the async thing ;-). So what happens when you don't use async, but sync? Is it safer? What's the difference in that practical matter? Will it *not* corrupt the file? But if the link goes down the transfer can't complete anyway.

Still it works reasonbly well tying two servers together this way. The only issue is that accounts need to be mapped from one host to the next. Groups and account names may differ. That is something to take account of. With SMB such things are not an issue. You read and write as a single user, more or less.

Thank you for your kind help. It kind of surprises me to come across a knowledgeable person who is actually nice. Most of the internet is so unhelpful particularly in the Linux world....

Which is why I want to be away from here ;-) (not meaning this forum as an isolated thing). I just want to be away... ;-).

x.
 
Old 12-21-2015, 07:29 AM   #7
TB0ne
LQ Guru
 
Registered: Jul 2003
Location: Birmingham, Alabama
Distribution: SuSE, RedHat, Slack,CentOS
Posts: 26,608

Rep: Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960
Quote:
Originally Posted by xenmaster View Post
Well in principle if I do any filetransfer that I know might get broken, I use rsync. Although usually HTTP downloads continue well, unless the browser that is doing the downloading deletes your files upon connection break-u p ;-). NFS is like this persistent filesystem that I can see from the exposure server. It works reasonably well although I didn't do much with it since that episode with the low transfer speeds. It's nice to be in a cloud interface and browse your files that way.

I forgot; the VPN used is over TCP which is why the traffic showed as TCP traffic even though it uses UDP. I would need a second VPN server instance to allow UDP. Not that I've ever seen it lower my transfer rates.

Still don't make me feel well about the async thing ;-). So what happens when you don't use async, but sync? Is it safer? What's the difference in that practical matter? Will it *not* corrupt the file? But if the link goes down the transfer can't complete anyway.
Async writes immediately; sync waits until the buffer/cache fills, THEN writes. That will usually explain a lower transfer rate. And if the transfer fails, no matter WHAT protocol/settings you're using, the file you're writing TO is still unusable anyway, so why does it matter?
Quote:
Still it works reasonbly well tying two servers together this way. The only issue is that accounts need to be mapped from one host to the next. Groups and account names may differ. That is something to take account of. With SMB such things are not an issue. You read and write as a single user, more or less.
You're now mentioning accounts, mappings, and SMB. Your final goal isn't clear, so it's hard to give any sort of decent answer. If you're doing things at one site, from rack-to-rack, you have far different options than if you're doing things over a slow WAN connection.

Again, async is faster...use it if your want faster NFS speeds, as well as using NFSv4 across the board. Aside from that, use SSHFS, which will not only be a bit faster, but easier to maintain.
 
Old 12-21-2015, 10:20 AM   #8
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Original Poster
Rep: Reputation: 0
Quote:
Originally Posted by TB0ne View Post
Async writes immediately; sync waits until the buffer/cache fills, THEN writes. That will usually explain a lower transfer rate. And if the transfer fails, no matter WHAT protocol/settings you're using, the file you're writing TO is still unusable anyway, so why does it matter?
I don't know, you mentioned it.

Quote:
You're now mentioning accounts, mappings, and SMB. Your final goal isn't clear, so it's hard to give any sort of decent answer. If you're doing things at one site, from rack-to-rack, you have far different options than if you're doing things over a slow WAN connection.
Accounts come into play with NFS. The source server at this point only has a single account and group, but if you NFS mount it on a different computer, that has different accounts/groups, the accounts don't match and you instantly have trouble with access rights .

But it's not rack-to-rack, it's across a slow WAN connection, as you say. There is an option of getting storage at the spot where the cloud server is located, so that would be rack-to-rack if I did that, it just costs extra money at this point.

I mentioned SMB because, weirdly enough, NFS doesn't use accounts, to connect, I guess it depends on access control based on server IP? I mean client IP. As long as my server (client) is on the access control list, I can mount the share, not needing any credentials. That is odd. Furthermore my local system (on the client) is allowed to write basically anything on the server, but the client is responsible for access control. That's very weird. They basically assume that the client is trusted and will do things in a dependable manner.

Odd assumptions that mean NFS is mostly meant to mount e.g. user home directories or data directories that form an integral part of the system that is doing the inclusion; in a sense, it is meant to be included by a system that is going to be "in charge" of the filesystem, and the NFS server is meant to be just a dumb filesystem server.

The use case of "two servers, each with their own user population" doesn't really fit in well with NFS. NFS is meant to mount home directories, not really media shares (as it would happen with SMB).

Why do I mention SMB? Because SMB DOES require a user account; and, as long as that user has access to the files (access control is done on the SMB server, not the SMB client) it will be able to read and write there, but also, on the /client/ all files will be mounted under a single predetermined user/group. So you get a wholly different story: you can mount it however you like, regardless of any users and permissions on the server side. It is not a "merging" or "blending" of two filesystems, but rather a real share getting mounted.

Windows has had a rather ephemeral sense of file ownership, in any case. Almost non-existent, and you never know what's when. But these are the two most prevalent options. Aside from that SSHFS, which I am still hesitant to use.

Quote:
Again, async is faster...use it if your want faster NFS speeds, as well as using NFSv4 across the board. Aside from that, use SSHFS, which will not only be a bit faster, but easier to maintain.
Right. Question: is it possible to use SSHFS as a regular fstab entry? I don't even have my.... VPN set up to be started at boot, I need to start it manually ;-). I never restart the system. I think there are more things that don't start automatically ;-). Don't really know how to do it... I hope it doesn't ever reboot .

Regards..

But so yeah at this point I am really seeking to discover how I can best mount a remote filesystem that is not part of the same LAN, keep it consistently mounted, and then don't run into trouble with the access rights.

I would synchronise the users and groups if I could, but the Synology NAS doesn't support custom groups below 65536. So on the cloud server I would have to create either an identical group at 65536 or run that mapping service that maps it to something else. There is all these issues to resolve. Then there is also the cloud service that uses its own users and won't have access to the share unless they are world readable, (which they are) and can't write to it unless it is part of the same group (that then needs to have write access on the server). It is all rather much of a complex, no matter how small.

At least rsync now stores using the proper group and the proper permissions. But it'd still need to setup that NFS mapper thing after I've determined what I want with it..... :-/.

So that's the bigger picture of what I'm doing. And the bigger picture of that is: what if the NFS server (Synology NAS) gains extra users that also write to it using their own user account on the NAS? Do I want to expose these files using the NFS mount, or only my own? If I make it writable from/through the cloud, -- which is easy to do, it just requires the right mapping on the NFS client -- do I want to be able to remove other people's files from the cloud interface? Will I give each of them an account on the cloud thing? Is it not better to expose the Synology interface directly on the web? These are all these sort of questions. I'm using OwnCloud and its file management interface is pretty good but not perfect. The NAS itself is not exposed, only its NFS share. And only through the VPN. So the cloud server also acts as a kind of buffer or defense.

It's like using two different cloud services at the same time on the same data, and it is a bit ugly.
 
Old 12-21-2015, 10:39 AM   #9
TB0ne
LQ Guru
 
Registered: Jul 2003
Location: Birmingham, Alabama
Distribution: SuSE, RedHat, Slack,CentOS
Posts: 26,608

Rep: Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960
Quote:
Originally Posted by xenmaster View Post
I don't know, you mentioned it.
I only mentioned that async is faster, and will corrupt a file if interrupted. Since the remote file will be bad no matter WHAT, it doesn't matter. You mentioned hesitation about this previously, by saying it was 'unreliable', which is why it was mentioned again. It still doesn't matter...async will give you faster write speeds, which is what your original question was about.
Quote:
Accounts come into play with NFS. The source server at this point only has a single account and group, but if you NFS mount it on a different computer, that has different accounts/groups, the accounts don't match and you instantly have trouble with access rights .
Not necessarily, depending on the rights on the mount-point. If the mount-point is 777, then EVERYONE on that remote system can access it.
Quote:
But it's not rack-to-rack, it's across a slow WAN connection, as you say. There is an option of getting storage at the spot where the cloud server is located, so that would be rack-to-rack if I did that, it just costs extra money at this point.

I mentioned SMB because, weirdly enough, NFS doesn't use accounts, to connect, I guess it depends on access control based on server IP? I mean client IP. As long as my server (client) is on the access control list, I can mount the share, not needing any credentials. That is odd. Furthermore my local system (on the client) is allowed to write basically anything on the server, but the client is responsible for access control. That's very weird. They basically assume that the client is trusted and will do things in a dependable manner.
Yep...typical of a VPS.
Quote:
Odd assumptions that mean NFS is mostly meant to mount e.g. user home directories or data directories that form an integral part of the system that is doing the inclusion; in a sense, it is meant to be included by a system that is going to be "in charge" of the filesystem, and the NFS server is meant to be just a dumb filesystem server.

The use case of "two servers, each with their own user population" doesn't really fit in well with NFS. NFS is meant to mount home directories, not really media shares (as it would happen with SMB).
Nope...NFS is meant to mount shares, not just 'home directories'. It doesn't CARE what you put on them..it mounts a remote disk as if it were local. There is no distinction with 'media shares'.
Quote:
Why do I mention SMB? Because SMB DOES require a user account; and, as long as that user has access to the files (access control is done on the SMB server, not the SMB client) it will be able to read and write there, but also, on the /client/ all files will be mounted under a single predetermined user/group. So you get a wholly different story: you can mount it however you like, regardless of any users and permissions on the server side. It is not a "merging" or "blending" of two filesystems, but rather a real share getting mounted.

Windows has had a rather ephemeral sense of file ownership, in any case. Almost non-existent, and you never know what's when. But these are the two most prevalent options.
Yes, but let's stick to your ORIGINAL QUESTION, which was regarding NFS speed. Use NFSv4 and async...you will then get better speed. SMB/CIFS gives you a whole other set of problems/issues to deal with, for no really good reason, unless you're doing things with Windows-only clients/servers, where you need Windows authentication/shares to be in effect. If you do, then please start another thread with your SMB related questions.
Quote:
Aside from that SSHFS, which I am still hesitant to use.

Right. Question: is it possible to use SSHFS as a regular fstab entry?
One of the first hits in Google for "mount sshfs in linux with fstab entry" is:
https://www.digitalocean.com/communi...stems-over-ssh

....which gives you the exact syntax. And I still have no idea why you're hesitant to use SSHFS. Not only is it simpler, but easier to administer and probably faster. It looks IDENTICAL to an NFS share, which means your remote file system is mounted locally to whatever mount-point you designate. That's it...the only difference is the protocol.
Quote:
I don't even have my.... VPN set up to be started at boot, I need to start it manually ;-). I never restart the system. I think there are more things that don't start automatically ;-). Don't really know how to do it... I hope it doesn't ever reboot .
All things that we can help with, so open another thread with further questions. When doing so, you will have to provide details.
Quote:
But so yeah at this point I am really seeking to discover how I can best mount a remote filesystem that is not part of the same LAN, keep it consistently mounted, and then don't run into trouble with the access rights.
...which goes back to NFS or SSHFS.
Quote:
I would synchronise the users and groups if I could, but the Synology NAS doesn't support custom groups below 65536. So on the cloud server I would have to create either an identical group at 65536 or run that mapping service that maps it to something else. There is all these issues to resolve. Then there is also the cloud service that uses its own users and won't have access to the share unless they are world readable, (which they are) and can't write to it unless it is part of the same group (that then needs to have write access on the server). It is all rather much of a complex, no matter how small.

At least rsync now stores using the proper group and the proper permissions. But it'd still need to setup that NFS mapper thing after I've determined what I want with it..... :-/.

So that's the bigger picture of what I'm doing. And the bigger picture of that is: what if the NFS server (Synology NAS) gains extra users that also write to it using their own user account on the NAS? Do I want to expose these files using the NFS mount, or only my own? If I make it writable from/through the cloud, -- which is easy to do, it just requires the right mapping on the NFS client -- do I want to be able to remove other people's files from the cloud interface? Will I give each of them an account on the cloud thing? Is it not better to expose the Synology interface directly on the web? These are all these sort of questions. I'm using OwnCloud and its file management interface is pretty good but not perfect. The NAS itself is not exposed, only its NFS share. And only through the VPN. So the cloud server also acts as a kind of buffer or defense.

It's like using two different cloud services at the same time on the same data, and it is a bit ugly.
Again, without relevant details, there's little we can say. Your original question was about NFS speed, and it's been addressed a few times now. Getting into account mappings, permissions, a Synology NAS appliance, and multiple servers/user accounts, and permissions mapping via VPN hasn't been in the scope of this. Again, all things we can help with, but details are needed about exactly what your environment is, and what your goals are.
 
Old 12-21-2015, 11:50 AM   #10
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Original Poster
Rep: Reputation: 0
Well that is called dialogue. A dialogue doesn't stay fixed on a single subject, but usually expands.

I have already mentioned that I might have been mistaken about the low speeds, at least, back then they were definitely the way I described, but right now, without having changed everything, I have the max transfer speed already (through the cloud "proxy"). So there is not much urgency anymore in that point, sorry.



Quote:
Originally Posted by TB0ne View Post
I only mentioned that async is faster, and will corrupt a file if interrupted. Since the remote file will be bad no matter WHAT, it doesn't matter. You mentioned hesitation about this previously, by saying it was 'unreliable', which is why it was mentioned again. It still doesn't matter...async will give you faster write speeds, which is what your original question was about.
But then why did you mention it, if it is of no consequence? Async is faster, and will corrupt a file if interrupted. Apparently that is relevant. I still don't know why, but you've left an indelible impression ;-).

My original question was really about read speeds, btw. So the writes happened on my local computer. The transfer was through HTTP across a link shared by NFS. I was getting very low transfer speeds (back then) and seemingly very high overhead, but I can hardly reproduce it at this point. So I don't know what was going on.

Quote:
Not necessarily, depending on the rights on the mount-point. If the mount-point is 777, then EVERYONE on that remote system can access it.
I believe this is not correct. In NFS from what I've seen the mount point doesn't make a difference. The actual access rights of the server are getting communicated. If I make it 777 then the files on the server are not going to all show up as 777. This is the case of SAMBA, yes, but not of NFS.

Quote:
Nope...NFS is meant to mount shares, not just 'home directories'. It doesn't CARE what you put on them..it mounts a remote disk as if it were local. There is no distinction with 'media shares'.
And yet all the access rights of all the individual files are getting communicated and are left intact from the server to the client, so there is a clear design intent.

Quote:
Yes, but let's stick to your ORIGINAL QUESTION, which was regarding NFS speed. Use NFSv4 and async...you will then get better speed. SMB/CIFS gives you a whole other set of problems/issues to deal with, for no really good reason, unless you're doing things with Windows-only clients/servers, where you need Windows authentication/shares to be in effect. If you do, then please start another thread with your SMB related questions.
Well or course, that is the question, isn't it? Whether there'd be any advantages to SMB, but I think it is less reliable and I was just asking whether it'd be any good option or not, thinking that it won't really. I have no specific SMB related questions at this point, I was just asking for an impression. Creating a new thread for every vague related idea or question you have doesn't work, that's like talking to another person each time you even slightly change the subject, or going into a different room each time you want to talk about something slightly different. It won't and wouldn't work.

And costs a lot of energy and rewards nothing.

Quote:
One of the first hits in Google for "mount sshfs in linux with fstab entry" is:
https://www.digitalocean.com/communi...stems-over-ssh
I was just curious whether you'd know. A computer can't think outside of the box ;-). A computer also cannot suggest anything based on actual intelligence.

Quote:
And I still have no idea why you're hesitant to use SSHFS. Not only is it simpler, but easier to administer and probably faster. It looks IDENTICAL to an NFS share, which means your remote file system is mounted locally to whatever mount-point you designate. That's it...the only difference is the protocol.
The difference is they call him king, the difference frightens me.

Quote:
All things that we can help with, so open another thread with further questions. When doing so, you will have to provide details.
Of course I know that, and I can find out myself as well (about startup scripts etc) but everything takes a lot of time and energy, and I don't have that, so I have to take things step by step. I was just alluding to these things because that in itself can mean a form of support arises, but not if you take it as meaning as that I require immediate support in the matter ;-).

Quote:
...which goes back to NFS or SSHFS.
Right now I already have NFS set up and I can easily attempt to use v4, but if I use SSHFS I will need to setup that as well, and it will probably create its own set of permission issues. Look, Linux is just filled with permission issues.

Just an example. I have an Ubuntu installed in a VirtualBox environment. I share a folder from the Windows host. I mount the folder, or at least access it, from the Ubuntu file manager. After this, through the file manager, I can write to the share. But what happens next, I try to extract files to the share using an archiving program, and it fails saying I don't have sufficient permissions to use the share as a target. So now I first have to extract the files to my desktop, and then copy them over. Completely senseless.

Typically with Linux (Windows these days as well) you CONSTANTLY run into permission errors. And if you happen to copy some file as root, then you have another problem, because the file will be owned by root even though it is located in your user folder. Then so, before your user can access it, root with have to chown it to your user. Bla. Bla. Bla.

The model is not user friendly. Worth thinking about though. How it could be improved or remodelled. One good step would be a filesystem flag that said "all files created in this directory will be owned by the directory's owner. Would solve a lot of problems.

Quote:
Again, without relevant details, there's little we can say. Your original question was about NFS speed, and it's been addressed a few times now. Getting into account mappings, permissions, a Synology NAS appliance, and multiple servers/user accounts, and permissions mapping via VPN hasn't been in the scope of this. Again, all things we can help with, but details are needed about exactly what your environment is, and what your goals are.
You simultaneously want to dive into new topics, while constantly saying that we can't. Make up your mind, you know. VPN = NFS. One thing leads to another. That's just the way it works (that's the way you can achieve results). You cannot achieve results be creating discontinuities in your thinking, by cutting one thing short and then hoping you can still start the next thing. That doesn't work, it falls flat on its face. In life as well, you have to move fluidly from one thing to the next because otherwise you'll crash down without support in the intervening phase. So yes I've had questions about related topics, and no I cannot go and open topics on everyone of them, as if they are all loose and disjunct items. I also cannot tell you my whole life story and then hope you'll fix it. There are certain things I don't know, so I ask questions about them or mention them in passing. Then it's up to you (or anyone) if you'd know something about it and if you'd want to hop on the trail.

But at this point, mostly, what I was asking was impressions, not fully formed answers. Because if you say "oh, that won't work in NFS" or "better not do that in SMB" then you've already given me a pointer. You've already given me a lead. That could very well be all the information I already need. You don't need to provide the full solution then. I'm just getting a bit of data, then, about what they call "best practices". As a computer user it is easy to come across technologies but it is harder to understand how to use them because the documentation writes usually don't tell you. When do people use NFS? When do people use SMB? What is the advantage of SSHFS? Those sort of questions are often harder to see answered. Only experienced people can answer them.

And you are one of those people, apparently, so I want to thank you for your attention and your answers regardless. They are at least some impressions about what I can use and what not. So thank you.

Btw, setting up SSHFS would mean seeing if my NAS supports it, etc, and then figuring out what would change. Then is the question of whether I want to only access it through VPN or directly between hosts. Etc etc etc. It gives more questions than I want to answer at this point and I still don't really see the benefits at this point. The NFS share currently is only used by one computer (IP). On the other hand, that SSHFS might be useful in mounting the NFS share remotely on this computer HERE. (However, installing win-sshfs gives issues on Windows 10). So SSHFS might be good for that but I don't think it is very good as a persistent mount across Linux platforms at this point.

Seems not very rich, etc. Might work well but I prefer a more solid mount, if that's the idea I'm getting from it.

Anyway. I will check into the speed thing later when more issues arise.

Regards.
 
Old 12-21-2015, 12:32 PM   #11
TB0ne
LQ Guru
 
Registered: Jul 2003
Location: Birmingham, Alabama
Distribution: SuSE, RedHat, Slack,CentOS
Posts: 26,608

Rep: Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960
Quote:
Originally Posted by xenmaster View Post
Well that is called dialogue. A dialogue doesn't stay fixed on a single subject, but usually expands.
Read the LQ Rules, please. Forums work based on threads...unrelated questions/topics confuse people who access forums as a knowledgebase.
Quote:
I have already mentioned that I might have been mistaken about the low speeds, at least, back then they were definitely the way I described, but right now, without having changed everything, I have the max transfer speed already (through the cloud "proxy"). So there is not much urgency anymore in that point, sorry.
Ok.
Quote:
But then why did you mention it, if it is of no consequence? Async is faster, and will corrupt a file if interrupted. Apparently that is relevant. I still don't know why, but you've left an indelible impression ;-).
And I don't know either, other than it seems you're missing the point. No matter WHAT you use (NFS, FTP, Samba, CIFS, SSHFS, etc.), if the file transfer is interrupted, the file on the remote side is not going to be worth anything. Async is faster, but file-corruption may only matter if you're doing something database related, or if you want that file 'live' on the remote system. In your case, as said several times, you were asking about speed...that is how you make it faster, and was only pointing something out. I believe you're blowing it out of proportion.
Quote:
My original question was really about read speeds, btw. So the writes happened on my local computer. The transfer was through HTTP across a link shared by NFS. I was getting very low transfer speeds (back then) and seemingly very high overhead, but I can hardly reproduce it at this point. So I don't know what was going on.
Think about what you were doing: NFS, over HTTP, over VPN...every time you add a layer, you're adding overhead.
Quote:
I believe this is not correct. In NFS from what I've seen the mount point doesn't make a difference. The actual access rights of the server are getting communicated. If I make it 777 then the files on the server are not going to all show up as 777. This is the case of SAMBA, yes, but not of NFS.
Nope, sorry. If the mount point on a remote server is 777, and the umask is set accordingly, anything you put in there inherits those rights. This only matters if you want user1 to ONLY see files belonging to user1...at that point, you need matching accounts/user ID's/groups, with the same numbers too, so things map correctly.
Quote:
And yet all the access rights of all the individual files are getting communicated and are left intact from the server to the client, so there is a clear design intent.
See above.
Quote:
Well or course, that is the question, isn't it? Whether there'd be any advantages to SMB, but I think it is less reliable and I was just asking whether it'd be any good option or not, thinking that it won't really. I have no specific SMB related questions at this point, I was just asking for an impression. Creating a new thread for every vague related idea or question you have doesn't work, that's like talking to another person each time you even slightly change the subject, or going into a different room each time you want to talk about something slightly different. It won't and wouldn't work. And costs a lot of energy and rewards nothing.
Again, read the LQ Rules. SMB will be just as reliable as any other protocol, but slower. And again, unless you really need Windows-specific permissions, there's no point to it. Further, forums DO work that way, this one has for many years. New questions = new threads.
Quote:
I was just curious whether you'd know. A computer can't think outside of the box ;-). A computer also cannot suggest anything based on actual intelligence.
Yes, and that is also amply documented. See the "Question Guidelines" link in my posting signature, which asks that people do basic research first, before posting a question.
Quote:
The difference is they call him king, the difference frightens me.
...and no idea what this means....
Quote:
Of course I know that, and I can find out myself as well (about startup scripts etc) but everything takes a lot of time and energy, and I don't have that, so I have to take things step by step. I was just alluding to these things because that in itself can mean a form of support arises, but not if you take it as meaning as that I require immediate support in the matter ;-).
Understood...please see "Question Guidelines"
Quote:
Right now I already have NFS set up and I can easily attempt to use v4, but if I use SSHFS I will need to setup that as well, and it will probably create its own set of permission issues. Look, Linux is just filled with permission issues.
...as is every other modern operating system. Windows has far more, which are more difficult to find. To use SSHFS, the ONLY thing you need to set up is SSH...that's it. Only the mount string changes.
Quote:
Just an example. I have an Ubuntu installed in a VirtualBox environment. I share a folder from the Windows host. I mount the folder, or at least access it, from the Ubuntu file manager. After this, through the file manager, I can write to the share. But what happens next, I try to extract files to the share using an archiving program, and it fails saying I don't have sufficient permissions to use the share as a target. So now I first have to extract the files to my desktop, and then copy them over. Completely senseless.

Typically with Linux (Windows these days as well) you CONSTANTLY run into permission errors. And if you happen to copy some file as root, then you have another problem, because the file will be owned by root even though it is located in your user folder. Then so, before your user can access it, root with have to chown it to your user. Bla. Bla. Bla.

The model is not user friendly. Worth thinking about though. How it could be improved or remodelled. One good step would be a filesystem flag that said "all files created in this directory will be owned by the directory's owner. Would solve a lot of problems.
Sorry, but this is totally wrong. There is a reason Linux is more secure, which is BECAUSE of permissions. You should NEVER use root unless it's absolutely necessary, and by doing so, you create problems. Linux permissions may be new to you, but they are NOT difficult, and are far easier than Windows to manage.
Quote:
You simultaneously want to dive into new topics, while constantly saying that we can't. Make up your mind, you know. VPN = NFS. One thing leads to another. That's just the way it works (that's the way you can achieve results). You cannot achieve results be creating discontinuities in your thinking, by cutting one thing short and then hoping you can still start the next thing. That doesn't work, it falls flat on its face. In life as well, you have to move fluidly from one thing to the next because otherwise you'll crash down without support in the intervening phase. So yes I've had questions about related topics, and no I cannot go and open topics on everyone of them, as if they are all loose and disjunct items. I also cannot tell you my whole life story and then hope you'll fix it. There are certain things I don't know, so I ask questions about them or mention them in passing. Then it's up to you (or anyone) if you'd know something about it and if you'd want to hop on the trail.

But at this point, mostly, what I was asking was impressions, not fully formed answers. Because if you say "oh, that won't work in NFS" or "better not do that in SMB" then you've already given me a pointer. You've already given me a lead. That could very well be all the information I already need. You don't need to provide the full solution then. I'm just getting a bit of data, then, about what they call "best practices". As a computer user it is easy to come across technologies but it is harder to understand how to use them because the documentation writes usually don't tell you. When do people use NFS? When do people use SMB? What is the advantage of SSHFS? Those sort of questions are often harder to see answered. Only experienced people can answer them.

And you are one of those people, apparently, so I want to thank you for your attention and your answers regardless. They are at least some impressions about what I can use and what not. So thank you.
Not going to re-hash forum rules again. Topics need to be focused.
Quote:
Btw, setting up SSHFS would mean seeing if my NAS supports it, etc, and then figuring out what would change. Then is the question of whether I want to only access it through VPN or directly between hosts. Etc etc etc. It gives more questions than I want to answer at this point and I still don't really see the benefits at this point. The NFS share currently is only used by one computer (IP). On the other hand, that SSHFS might be useful in mounting the NFS share remotely on this computer HERE. (However, installing win-sshfs gives issues on Windows 10). So SSHFS might be good for that but I don't think it is very good as a persistent mount across Linux platforms at this point.

Seems not very rich, etc. Might work well but I prefer a more solid mount, if that's the idea I'm getting from it.
Again, wrong. SSHFS works quite well for persistent mounts, and has for a while now. I use it in production systems right now. Again, SSHFS ONLY NEEDS SSH, which you already have on that NAS. SSHFS is also covered on the Synology forums, where many users report success with it. And again, as long as you have an SSH route between hosts, it doesn't matter if it's running through VPN or direct...it just works.
 
Old 12-21-2015, 12:43 PM   #12
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Original Poster
Rep: Reputation: 0
Why are you so upset with me not jumping on the SSHFS bandwagon straight away?
 
Old 12-21-2015, 01:03 PM   #13
TB0ne
LQ Guru
 
Registered: Jul 2003
Location: Birmingham, Alabama
Distribution: SuSE, RedHat, Slack,CentOS
Posts: 26,608

Rep: Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960
Quote:
Originally Posted by xenmaster View Post
Why are you so upset with me not jumping on the SSHFS bandwagon straight away?
Not upset in any way, but you appear to be missing the point, and seeing complications where none exist. At any rate, I'm done...good luck to you.
 
Old 12-21-2015, 03:08 PM   #14
xenmaster
Member
 
Registered: Sep 2005
Posts: 32

Original Poster
Rep: Reputation: 0
Quote:
Originally Posted by TB0ne View Post
Read the LQ Rules, please. Forums work based on threads...unrelated questions/topics confuse people who access forums as a knowledgebase.
That is outright selfishness and serves to disenfranchise the question asker in favor of that unknown crowd who is, to you, more important than the person you are actually talking to. So you are actually not trying to help me, you're just trying to build a knowledge base and you just pretend to be in it for helping actual people.

You can be an ass all you want and think that people need to comply with your wishes, but you still can't turn nature around to do your bidding. And as you are not in it for the person who is asking the question, apparently you do not care, you simply do not care that it does not work for that person.

If this is a place to get answers, then your stance DOES NOT WORK. We people do not work that way. But you do not care about people, you only care about your knowledge base.

It's the same on many other forums, and I have no need at all to read your copy and pasted forum rules, because there is nothing new in it. You are all greedy selfish bastards.

I was recently on an ArsTechnica forum and I asked a very simple question. Yet people refused to help me because they thought that if they helped me do the thing I wanted, that it would be bad advice for all those other people that might be reading the thread since they thought that what I was doing was the wrong way to do it or much rather the wrong thing to do. And then those people said quite literally "this is not a place for [Xen] to get his odd questions answered, this is a discussion forum". So their goal was not to provide me with any answers, but rather to debate my choice (and try to get me in line again).

The result of that debate then being (hopefully, to them) that the clear result of the discussion would be that what I wanted (for my personal situation) was the wrong thing to want. And they hoped the thread would be educational in telling people (in general) what solution to choose regardless of personal circumstances. Since my situation is rather different, than the average, or the general, their solutions obviously won't work for me, but still they are trying to sell them to me. And they don't care about me, since they want their answers to be applicable to the general crowd, as if they want to compete with that horror of a story called StackExchange.

I believe the climate on the internet has become much worse since that StackExchange network rose to dominance. Suddenly everyone is concerned with building the best knowledgebase and using their users for that. WE are just being USED for someone else's goal.

People are not sincerely trying to answer the question or help a person. People are using questions as means of giving answers that will apply to the general audience -- precisely what StackExchange set out to do -- in order to become a knowledge base that plays well with Google et. al. Obviously, if that is your goal, then any deviance in a forum thread, any real sense of dialogue, anything that does not contribute to a searchable knowledgebase (but that would actually help people with their problems) is a detriment to that goal.

And those are then your precious "rules". The rules are: you are just being used here, we are in it for the money, and we don't really care about your situation or your problems, as long as we get what we want: which is better search rankings.

I'm sorry but I'm not interested in being an unpaid employee in building your assets. I was in it for a mutually beneficial transaction in which a human being helps another human being achieve his goal and in the process both learn something. Forums are a public venue (or avenue) and of course it helps if content is searchable, but you disregard the fact that people are also interested in reading discourse and this is what forums are. One earlier time on some other forum I was chastisted for posting experience reports on a forum that (as a whole) was meant for "asking questions". The relevant subforum.... hadn't seen any activity for over 10 months.... and yet I was not allowed to post about what I had sought out to do and what I had learned --- as if that would not be of interest to people visiting the forum because it had no "question-answer" format.

A forum is NOT a fixed format for knowledge building. That is what you are trying to turn it into. And moderators are usually trying to prevent people from doing what they would naturally do. A forum is a venue for public discourse. Discourse is a free-flowing thing. So you are basically trying to prevent what is natural, from occuring, because you have ulterior motives. What you want doesn't work for the people who actually come and visit your forum, but since you are not in it for them, but rather for yourself (or your group/company) and what you want is at odds with what they want (your visitors) you don't really care about the fact that it doesn't actually work (for the actual people).

It's not the forum that doesn't work. The forum works in lieu of your rules. The forum works regardless of what you are trying to turn it into. Not because of what you are trying to turn it into. Regardless of that. It's the rules that don't work, because the rules don't comply with what a forum is, and the rules seek to combat the natural tendencies that people have on a forum.

It is you that misunderstands the forum, not me.

It is you that is not clear on when something works, and for whom, not me.

Quote:
Yes, and that is also amply documented. See the "Question Guidelines" link in my posting signature, which asks that people do basic research first, before posting a question.
Quote:
Understood...please see "Question Guidelines"
Do you really think I'm this dumb, or would it rather be that you (or you and your people) are this dumb?

You have copy and pasted forum rules and forum guidelines from some other forum, and basically every forum on the web states almost the exact same rules and guidelines, and still you think that you are doing something new, that what you are saying has any sense of value.

You really think that a person of my stature has never read those stupid rules before? That it is something new and refreshing you are alluding to?

How dumb.

How utterly dumb.

Just because a person disagrees with your copy and pasted rules doesn't mean he is unaware of them.

Just because your rules don't work for a person, and ask a person to become some unpaid volunteer or employee, and that is in disregard of what people actually come here for... doesn't mean the person is too stupid to understand them, it is because you are too stupid to understand them.

You don't understand that they work against people and seek to disenfranchise people. Who actually come here. To ask questions.

Time and time again you people try to mold people according to your wishes, try to make them do what coincides with your stated goals. They are not your visitor's goals, so you try to coerce them into it.

Constantly we are being coerced, forced. We want to do a certain thing, but it is not allowed, because it does not build towards that knowledge base that the forum can never be. By its very definition a forum is a temporal thing, and you do not use it to build eternal knowledge. That is not a goal that agrees with what a forum is.

So your concern is with the "there and then" not the "here and now". But my concern is with the here and now, or I would not be here asking a question now.

And you want to format our dialogue in such a way that it benefits the there and then while sacrificing the here and now. And I protest against that. I object to that. Of course I do. You are harming me and my interests. I'm not here for you. I thought you were here for me, but apparently you think it is the reverse.

You think that I'm here for you and that you are not here for me. You think that you are here for yourself and that I am here for you as well.

The complete opposite of what a question is. Invent your own questions then. Might get boring though. When the fašade of helping people falls away.

Quote:
Not going to re-hash forum rules again. Topics need to be focused.
They need to be focused for you, not for your clientele. What you do is at odds with what people want.

You are trying to bully people into your wishes.

And I'm not having that.

I even have to explain this stuff to you.

Incredible really.

You think that what people want harms your interests, and you are even mistaken in that. But as long as you keep thinking this way, you will never see it.

Let me tell you at least that StackExchange is the most horrid place I have ever seen on the internet, the vileness, the hatred, the criticism, the condemnation, the judgment, it is worse than the vitriol in e.g. a game of League of Legends (not that I've played it). At least in LoL you can still have fun. It's just incredible that I need to educate people about this.

Seeking to be like StackExchange is NOT a good way to be. No matter if you existed way before they did. I talk of LQ.org. As you can see my account is also from 2005. You've been here longer, yes, and you have a few thousand posts more to your name ;-). Still, using a forum to build a 'knowledge base' is a harmful thing to do. You are using people and that is harmful! I'm not saying this is a bad forum. I haven't been here much. I come here because it is a more general place and people are less inclined to be particular about what you want.

Like if you go to an OpenSUSE forum they are going to be hateful if you try to do things the Debian/Ubuntu way. If you go to a Kubuntu forum they are going to be hateful if you try to do things the OpenSUSE way ;-). Jeez.

But no, if I don't follow the "rules" it is not because I am dumb, it is because you are dumb.

You don't have to push these rules in my face and you don't have to think I would not be aware of what the general line of thought is that is going to be stated in those rules.

You also don't have to think that I haven't read those "how to ask questions the smart way" documents of yore. The ones that originated on news groups or mailing lists. The ones from the Unix gurus. But by all means, don't coerce me into what you want. I'm not coercing you either.

I asked a voluntary question in receipt of a voluntary answer. You only have to answer if you want to. It is all about freedom. You are not obligated to answer my question. Many forum regulars feel that way; that they have an obligation. Then they become angry if the question is "wasting their time".

Then they start to point at "GIYF" and "RTFM". You have done the same albeit in a nicer way. If a person asks a question you are not compelled to answer it! If a person has a slight interest in obtaining more information, he or she can indicate this. This doesn't mean he or she is obligated to do "research" either! You people constantly think this is a game of obligations; you are obligated to answer, which means I am obligated to form a good and full question.

Forum 101 really, jeez.

And this lack of common sense, and this lack of basic decency arises because there are ulterior motives and people are often not (solely) in it for helping the person who asks the question, no they have the value of their forum content in mind according to some oddball definition of what makes a forum base valuable.

You say it has worked; by that I mean it has worked to raise the forum value according to that oddball definition of worth. It doesn't mean, or doesn't necessarily mean, it has worked to help people get the answers they want.

Currently, today, it is not entirely working. That is a fact.

I am pointing you to the fact that right here and now, it is not working.

You are also becoming angry because I disagree with you.

I reject your solution and now you are becoming angry. I haven't read your response yet above this one.

Apparently you are becoming angry because my discussion and rejection does not comply with the "fixed format" you want forum threads to have?

Is it because I deny your expertise and I claim to know better (? about myself)/?.

At first you were very friendly but with each passing post you become more angry and more belligerent. Apparently I'm doing something wrong but I don't know what it is ;-).

Ah, maybe I haven't read the forum rules, that would be it. Or, I don't agree with them. My nature doesn't agree with them. Human nature doesn't agree with them.

Good cause for anger yes.

If something ain't working for me, then it ain't working for me. No amount of rules is going to change that. The thing you want me to do is not going to happen unless it works for me. Yes, for me.

And hopefully it works for you as well, but that depends on what you think your stated goals are.

I was hoping this would be a 'mutually beneficial transaction' but currently you are:

- refuting anything I say based on really not much
- trying to force me to agree that SSHFS is the best solution for me

You are:
- throwing in authoritative conditions to make your point ("I use these in production environments, so it works (for everyone)")
- not yet calling me passive-aggressive, but soon you will be ;-) :P.

- Probably going to threaten again with one thing or the other instead of engaging as equals.

Going to agree with me that what you've done has brought this thread VERY much off course ;-).

I'm sorry, but if people offend me this much, I just have to make note of it :P :P :P.

You've not let a person be, trying to force him into your way and he has objected.

You have sought to make the forum more efficient, and it has become less so. That doesn't work.

Soon you or someone is probably going to start deleting posts to bring back the efficiency in some panicky attempt to restore order.

That means you would throw away a whole lot of what a person has written which is a great offense. That's just an insult.

That means you'd be insulting your customers, because after all, they're not really your customers. I don't know who is, because there are no ads.

But your investment is in building this "KB" and I am simply not in agreement with that in the way you are doing that.

Because you are not allowing me to speak my mind and you are treating me as your subject. As subtle as you've done it in the beginning, in this post you've told me to "read the rules" at least 4 times.

As if I'd be so dumb not to understand what they say. That is an insult in itself.

And because you don't like the fact that I don't immediately agree with your suggestion or solution, you insult me even more. You even frame my disagreement in not complying with the rules, because you had no problem suggesting SSHFS as long as I hadn't voiced a disagreement yet.

Well.

I'm not sure what else you can ask:

- I gave you feedback on your own treatment of me which means I am no longer passive-aggressive, now I'm aggressive :P.
- I told you that the forum will do fine or even better if you actually allow people to ask the questions they want /because people like discourse/. And it can give them new ideas as well.
- I suggested that there is no need to treat people like this, at least not me.
- I told you that you can easily change this and it is not a fault of your own
- I suggested that it'd be more respectful to not coerce people.

- I made fun of you and everyone
- I made fun of myself.

What more do you want? :P.

Haven't I been an excellent forum member?

Next week I'll get a reward for not getting instantly banned after voicing my concerns.

Perhaps.

But it is pretty clear that your "read the rules" are way more off-topic than anything I've said. About NFS. And you spin a debate that is even more 'offtopic'.

Except that it is on the topic of what you've wrote.

So in a sense you are the one to blame for any off-topic-ness because as is natural to a forum, debate can diverge /before coming back to the topic at hand/ and there is often much more to a topic than can be described in a single headline. You've started side-lining the discussion with your forum rules.

And I was *just* expanding the topic a bit because obviously speed issues are contingent on architectural issues. So it is rather obvious that I would /tell more of what the architecture is/ as I go deeper into /why the speed issues arise/. And then you keep saying "other thread, other thread". And what am I supposed to write in that other thread? I have no question outside of this thread.

Because there is nothing that can have a clear-cut answer. You've suggested async and I've said numerous times now that the speed issues are gone.

You even berate me of running NFS through VPN over HTTP but my bandwidth is 500KB/s and I am currently getting 500KB/s. So I apologize for posting this thread in the sense that apparently I hadn't tested it enough. Apparently the problems didn't arise because of NFS and I don't know what did it. So using SSHFS currently has no benefit, and neither does using v4.

Apparently you want me to accept your answer and immediately accept your advice. Sorry, it doesn't work that way.

Quote:
Again, wrong. SSHFS works quite well for persistent mounts, and has for a while now. I use it in production systems right now. Again, SSHFS ONLY NEEDS SSH, which you already have on that NAS. SSHFS is also covered on the Synology forums, where many users report success with it. And again, as long as you have an SSH route between hosts, it doesn't matter if it's running through VPN or direct...it just works.
Why so fucking angry? I once had a "friend" who advized me to get a camera without a battery pack but using penlites, non-rechargeable. I told him I didn't feel for that and he became angry.

I don't feel for SSHFS, what's to give?

What does it matter to you? Or to me? I apparently don't need it to solve the problem and I don't like it architecturally. I would use it on my remote Windows desktop if I can get it to work but I'm worried I'm gonna mess up the harddisk because it will be so easy to modify and delete stuff while getting it there takes so long. I just don't like it. Clear.

I don't like SSHFS as a persistent mount. My SSH sessions are also not persistent and often break. Normally happens a few times a day that my connection gets closed. At the same time, the SSH between the two servers that I mentioned often sticks around for days.

Quote:
Not upset in any way, but you appear to be missing the point, and seeing complications where none exist. At any rate, I'm done...good luck to you.
Yeah, the point is that I need to take your advice no matter who I am or what my wishes are, and that if I do not, you become angry.

Well good riddance then. Hostile person.
 
Old 12-22-2015, 07:34 AM   #15
TB0ne
LQ Guru
 
Registered: Jul 2003
Location: Birmingham, Alabama
Distribution: SuSE, RedHat, Slack,CentOS
Posts: 26,608

Rep: Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960Reputation: 7960
Wow, really??? We're all, as you say, "Greedy, selfish bastards". Funny, you came here for help and answers, and GOT THEM...FREE...from the same "greedy, selfish bastards" you're ranting about. And NO ONE HERE is an employee...ALL of us are volunteers; or as you say 'unpaid employees'.

Can't be bothered to read/adhere to the forum rules? That's up to you, which explains why you have gone from forum to forum. You're not a special, precious little snowflake; rules apply to you the same as they do to everyone else. Don't like it? Then NOTHING is preventing you from starting your VERY OWN FORUM, with your own rules.

Quite sorry I answered your questions and helped you, and that you couldn't seem to understand the points. As said, good luck to you...with your attitude, you will most definitely need it.
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
LXer: What Inefficient Airline Boarding Procedures Have To Do With Net Neutrality LXer Syndicated Linux News 0 05-03-2014 07:00 AM
Aren't text-based interfaces inherently inefficient? (like procfs) hydraMax Linux - General 16 08-02-2012 12:05 AM
Pygame slow or just programmers are inefficient? General Programming 5 03-23-2010 04:06 AM
My system feels inefficient in its use of CPU power Changes Linux - General 7 10-06-2009 10:49 PM
rsync - inefficient behaviour? Unclesmiff Linux - Software 4 01-02-2008 11:13 AM

LinuxQuestions.org > Forums > Linux Forums > Linux - Networking

All times are GMT -5. The time now is 03:22 PM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Open Source Consulting | Domain Registration