[ViX_Misc] AutoFS bug?

[Piranha]

I have had a long-standing problem getting network shares to work consistently between my two TM Twin boxes. After a reflash (with or without settings restore) mounting the HDDs as shares using AutoFS works a treat, typically for a few days, then when trying to access the shares through the media list and Yellow button (Location) I get the dreaded 'directory media/autofs/XYZ does not exist' message.

I have just finished experimenting as follows:

1. Logged into router and confirmed that both boxes have static IPs below the DHCP start range.
2. The router sees both boxes with the correct IPs.
3. Each box shows the other box' HDD in the Network Mount Manager (Mount points management) correctly.
4. I changed 'Mount using' from 'AutoFS' to 'Enigma2 (mount using enigma2)' on both boxes and the shares then worked instantly!
5. Tried again using AutoFS, rebooted and got the '......does not exist' message again.
6. Back to mount using Enigma2 - worked instantly - and seems a lot faster access time than when AutoFS is working too!

I conclude that the automount has a bug in it, which has been in the image for a long time now.

I suggest that anyone else having the same issue tries using the Enigma2 mount option and reports their findings, maybe the ViX team will be able to have a look into this problem.

 

[tuneboy2005]

Check the "NFS issue" topic. I bet it has something to do with my problems and others too. Mounts that are dissapearing.

 

[Piranha]

So much for my 'Eureka' moment a post #1 above - the shares still disappear no matter which way they are mounted i.e. AutoFS, FSTAB or Enigma2 - although Enigma2 method gives much faster access to the network drive than AutoFS for some reason!
Both boxes are wired to the router and have static IPs. Via ViX Network Browser, each box is found and expands to show the HDD mounts. I have setup the shares literally scores of times and they work instantly, but only for a period varying between a few hours and a few days.
I am now going to plod through the debug logs of both boxes and see if there are any clues as to what happens when trying to navigate to the shared HDD, although I am not hopeful.

Final point - just checked again from one box and a mount that said 'media/net/LOUNGE does not exist' a few minutes ago now works! Wish I could figure out what causes the inconsistency!

Will report my findings in this thread.

 

[Piranha]

Interesting find in the Debug log for the LOUNGE box - following a reboot or E2 restart, in the boot sequence there is the line:

new Network Mount BEDROOM -> /media/net/BEDROOM

followed a few lines later by:

command ['umount - fl /media/net/BEDROOM', 'sleep 2']

Can't find any clues as to where the 'umount' command comes from!

I will keep experimenting with network shares in Apollo 165 for the moment and then see if the new image series fares any better, though I doubt there are any changes that would affect this component.

 

[tuneboy2005]

Like i said, to test i have switched to the Blackhole image with same mounts and everything is running fine for 5 days now, no mounts that are dissapearing. i will flash the new vix image to see if problems occures again.

 

[birdman]

So much for my 'Eureka' moment a post #1 above - the shares still disappear no matter which way they are mounted i.e. AutoFS, FSTAB or Enigma2 - although Enigma2 method gives much faster access to the network drive than AutoFS for some reason!

autofs would need to mount the file system before accessing it, which might take a second or so. After that the transfer rates should be the same (although what Enigma2 is if it isn't an fstab entry bemuses me).

One quick thought is that the NFS connexion on Vix uses tcp. Just wondering whether Blackhole uses udp. Could be that something somewhere tries to chop an inactive tcp connexion....

 

[Piranha]

autofs would need to mount the file system before accessing it, which might take a second or so. After that the transfer rates should be the same (although what Enigma2 is if it isn't an fstab entry bemuses me).

One quick thought is that the NFS connexion on Vix uses tcp. Just wondering whether Blackhole uses udp. Could be that something somewhere tries to chop an inactive tcp connexion....

Thanks for your interest birdman - I'm afraid that a lot of this stuff is black magic to me!

One thing I can do, like tuneboy 2005, is run a box with a different image for a while, just to see if the network mounts work as they appear to be doing in BH for him. I'll try HDF or ATV and see what happens.