Posts by RoccoJones

There's definite a problem when I mount an NFS on LE. It was bad when I was running LE 7, but now that I've gone to 8, it's much worse. It manifests itself as never recording more than 15 minutes of OTA programing. However, I can demonstrate it 100% of the time w/something like

dd if=/dev/zero of=/storage/2g bs=1024 count=2097152

It usually fails before 100Mb is written w/

Quote

dd: writing '2g': Input/output error
68274+0 records in
68273+0 records out
69911552 bytes (66.7MB) copied, 16.769586 seconds, 4.0MB/s

Another interesting key to the problem is that the mount has to be done from /flash/cmdline.txt. If it's done from the command-line after boot, the full 2Gb (or any amount you like) is written.

I usually use a Raspberry Pi. B+ as the NFS server, but I also tested w/a Ubuntu system as the server and they both failed the same way. So the problem is definitely w/the NFS client.

LE is running on an R.Pi 3. My cmndline.txt

Quote

boot=UUID=A5AD-A5ED disk=NFS=192.168.1.105:/mnt/mainhdd/dospace/,wsize=65536,retrans=5 ip=192.168.1.103:192.168.105.1:192.168.1.1:255.255.255.0:DosPace:eth0:off:208.67.220.220:208.67.222.222 quiet

If you want me to diagnose anything else, it should be before next weekend, because I'm probably going to see if OpenELEC work better.

Well, I guess I'll never know

And when you do, you don't have to spend a lot of time working on the retrans, because bumping mine up to 5 doesn't appear to have made a lick of difference.

While you're working on that, I'd like to try a stop-gap measure like increasing the disk write cache, at least for that mount. How can I do that?

I have a problem w/mounting an NFS then ends in an i/o error. If I run

dd if=/dev/zero of=2g bs=1024 count=2097152

then it quits after 100Mb or so. Here's my latest attempt:

Quote

dd: writing '2g': Input/output error
114898+0 records in
114897+0 records out
117654528 bytes (112.2MB) copied, 28.075459 seconds, 4.0MB/s
[END QUOTE]

I might have blamed it on the server, but if I mount the same thing from a Ubuntu laptop, I'm able to copy 10Gb to it w/out problem.

Here're the options:
192.168.1.105:/mnt/mainhdd/dospace/ on /storage type nfs (rw,noatime,vers=3,rsize=32768,wsize=65536,namlen=255,soft,nolock,proto=tcp,port=2049,timeo=3,retrans=2,sec=sys,local_lock=all,addr=192.168.1.105)

Display More

It looks like the answer to my idea about the root problem on the server side from a couple of weeks ago was 'yes'. I decided to look at it from the server side, and implement the no_root_squash (or whatever) option and finally, I got my Pi to start w/NFS.

N/m about that "root="--that was just a bad copy/paste.

Assuming you mean

boot=UUID=A5AD-A5ED disk=NFS=192.168.1.105:/mnt/mainhdd break=load_splash debugging ip=dhcp

I gave

mount -t nfs 192.168.1.105:/mnt/mainhdd /sysroot

it just timed out this time. At least it probably would have, if I didn't think 5 minutes was enough and just cancelled it.

Quote from milhouse

There is no root= option,

Are you sure? I'm pretty sure that "root=" came w/the distro.

So what exactly should my cmdline.txt be now? I've changed parameters so many times I don't know what it should be now. If I'm still debugging, what do I do when I get the command-line?

Very well,

[code]
root=UUID=A5AD-A5ED disk=NFS=192.168.1.105:/mnt/mainhdd/ break=load_splash debugging ip=192.168.1.103:192.168.105.1:192.168.1.1:255.255.255.0:DosPace:eth0:off:208.67.220.220:208.67.222.222

[/code]
That got me my shell, and it then let me mount the NFS from the command-line. So I've managed to get that far. How do I go about turning that into something that will mount on startup?

Same problem: black screen, nothing going on. I wouldn't swear on my life that all the parameters are right, though:

root=UUID=A5AD-A5ED disk=NFS=192.168.1.105:/mnt/mainhdd/ break=load_splash quiet ip=192.168.1.103:192.168.105.1:192.168.1.1:255.255.255.0:DosPace:eth0:off:208.67.220.220:208.67.222.222

For instance, there shouldn't have been any need to give the gateway since the NFS server is on the same subnet, but I didn't know what else to know.

That definitely had an effect, but now it just froze the Pi w/a blank screen. Attempts to ssh in were just met w/"Connection refused". This was still w/the "break=load_splash" btw. Could this be related to the root problems I mentioned earlier, where a successfully mounted NFS couldn't act as root?
If doing it through were a long-term workaround, it would present difficulties as I need a static IP to locate the Pi. My router's server is supposed to be able to assign specified addresses to specified MAC addresses, but it doesn't seem to work.

Then maybe someone wants to weigh in, because someone doesn't understand something: Raspberry Pi • View topic - Can't mount NFS through cmdline.txt

Ok, I followed your debugging advice, and in the end, the reason it seems to fail is that the network doesn't come up.

Once I re-established the normal disk, I logged in and pinged the NFS server.

So clearly this is a problem of the network being down when the mount is attempted.

That's what I said. Then the other guy said:

Quote

Same answer as before, NFS or NTFS your mount command is non-standard so there's probably some funky LibreElec script doing it.
(END QUOTE)

So they seem confident that ntfs-3g is necessary for this. In any case, I'm told very little about why it fails when I start, and I wouldn't know where to go to find more.

I just went and asked the Raspberry Pi guys about this and got this answer:

Quote

You also need the ntfs-3g driver built into their Linux kernel or as a kernel module to support mounting NTFS.
(END QUOTE)

Does that give any more insight into what the problem is? The guy went onto say that the mount command is non-standard and that it must be a LibreELEC script managing it.

What you are describing is exactly what was going on before my hdd melted down.
Nah. It's doing the exact same thing, using any and all 5 frequencies for each of channel 13 and 17 from North American television frequencies - Wikipedia. At least it's exactly the same if I use the "ATSC" "Delivery system". If I use the "ATSCMH" "Delivery system" (whatever that is) it just sticks in the "PENDING" state ad infinitum. Of course, I have no idea what the hell I'm doing.
It doesn't seem like tvheadend is getting a lot of attention. What's the next best alternative? From what I've seen, it's impossible to get a proper consumer grade PVR system for love or money.
[hr]
...actually, another thing that 4.0 afforded me are those pre-selected muxes, which everything I've read told me to avoid. Doing so created all those muxes w/those elusive frequencies which the tvheadend guys wouldn't tell me. Now I can record OTA channels on Kodi
So even thought I seem to have got it working, my question still stands about a better alternative to tvheadend, if anyone knows of any.
[hr]

So how do I troubleshoot this? From what it seems, there's no log when this happens and the feedback I get is minimal.

Following Mounting network shares - OpenELEC it mounted ok. But because I didn't fix it w/the options, there was a problem accessing any files or directories that didn't have global permission. The 2 roots are, of course, different. Could that be the problem?

Quote from lrusak

If you just change the "disk" value it will work, don't touch the "boot" value

Code

disk=NFS=192.168.x.xxx:/path/

Didn't work. My cmdline.txt is:

boot=UUID=A5AD-A5ED disk=NFS=192.168.1.105:/mnt/mainhdd/ quiet

(boot field unchanged) When I tried to restart, it just said it couldn't boot from that. When I try executing:

sudo mount -t nfs 192.168.1.105:/mnt/mainhdd/ nfsclient/

on Ubuntu or even the Raspbian server that 192.168.1.105 is, they run fine.