My observations ( for people that have indexing and connection issues )

Question

Since cisco takes an awful lot of time doing anything about these problems i decided to investigate the box myself today.
I followed advice from various threads and did some digging aroudn with network traffic snoopers and some other tools.
First ( and this is buried somewhere deep in the documentation ) you can NOT have ANY file or folder name that is longer than 32 characters. This is a limitation of the stupid linux they are running on ! If you have a file with longer name or foldername it may screw up the indexing.
Second : file and path names can only have alphanumerical characters and numbers and spaces in them. DO NOT USE ANY other character. it messes up the indexing  I had folders named -= folder 1 =-. they never got indexed. as soon as i removed teh  -= and =- the indexing kicked in .... This is again a twonkyvision / loonix problem
The box has tremendous problems with ghost devices on the network. I have on my network at home : 4 pc's running XP ( some pro some home edition )  2 laptops with Vista. One box with Win7. One Windows home Server. One linksys Skype phone , One Roku soundbridge, one Dlink DNS323 , one ADS network drive , one Simpletech Simplestor , A HP color laserjet , a Hp 6250 , 7650 and some other HP network printer. Thru the wireless link my PDA ( iPaq) and iPhone connect once in a while too. My blu ray player is also hooked up (netflix streaming). And then there are various experimental systems ( i am an electronics engineer and i have built some gadgets that are network connected that allow me to remotely turn on lights monitor temperature etc. )
Now , the NMH does not detect correctly most of the devices. It keeps on trying to feed information to the printers ... it also tries feeding information to other NAS devices as well as to the Windows Homeserver ... It falsely identifies one of the printers as a dlink ethernet connected dvd player ...
It also has problems with devices that use static ip addresses on the network ( i set up the printers and other NAS devices with hardcoded ip addresses. so no dhcp )
So here is what i did : go to twonky configuration ( port 9000 see tony's article )
Step 1 : Yank out the ethernet cable to ANY OTHER DEVICE except the pc you are working on , your router and the NMH <- this is important
step 2 : Hit the button to erase all the devices it discovered.  (reset list)
Step 3 : hit the SAVE button
step 4: UNCHECK the box next to the 'reset list' button
step 5: hit the SAVE button
Now , on the left hand side click on Maintenance
Click on ALL checkboxes under Log level. they should ALL be checked
Hit the Clear Logs button
Hit Save changes
Hit RESTART server. you wil get a file not foound error page. you will see in your browsers titlebar that the url changes to an ip address with some text behind it. remove all that text and key in :9000 and hit return. it will take you back to tonky. ( i don't know how important this step is , but i did not go in throught the device name , i used the ip address from this point on. normally it should not matter but you never know ( i have a suspicion i will explain later )
Write down this IP address. it is usefull to know.
Now since we are back in wonkyvision ( stupid half baked program ) Go back to the maintenance screen.
now hit the rebuild databse button.
You should hear disk activity now.
Hit the Show log  button once. the log files should open.
you can refresh this screen by hitting the reload button in your browser.
you should see messages fly by like
21:14:40:317 LOG_SYSTEM:fsmon_add_watch inotify_add_watch (12, /share/media/photos/Moorea May 1998) returns 2045
21:14:40:317 LOG_DB:watch wd=2045 added on /share/media/photos/Moorea May 1998
21:14:40:379 LOG_DB:upnp_folder_watch_add_dir /share/media/photos/Moorea May 1998/New Folder
thnis means it is probin the entire directory structure and adding files and paths.
Let it run for a while. hitting refresh on your browser once in a while. it took a couple of hours on mine ( 76000+ pictures ... ) and a couple of hundred songs + some videos.
once disc activity ceases : go back to twonky port 9000 , go in to maintenance and hit Clear Logs button.
HitSave button
Hit restart server. you will again get an error page. get rid of the rubbish behind the ip address.and key in :9000
go to the clients and security page.
Make sure Automatic discovery is still turned OFF ( if it is on you will see in the logbooks that it attempts several times a second to connecto to anything it can find. Since the detection process is flawed it bombs out. this may overload the poor cpu in the NMH... )
Hit the reset list once more
hit save
Now turn automatic discovery on and hit Save.
go back to maintenacne and hit restart server. again on the error page : erase the garbage after the ip address and go back to port 9000.
if you now go back to client/sharing page you should see 2 possibly 3 device. one is your router , one is the pc you are working on, and the last one is the same ip address as you see in the browser. ( the ipaddress you are using to talk to the nmh )
make sure all the checkboxes before these devices are checked. and hit the save button once more.
at this point i unchecekd the automatic discovery and hit save once more. i go back to maintenance and hit restart server for the last time.
At this point the nmh restarted twonky and immediately there was a ton of disc activity. i opened the normal nmh user interface and lo and behold : the green spinning arrow started to move and progress was going forward. it increased 1% roughly ever 10 seconds or so. when it finally hit 100% everything was there. as it should be.
Now. speculation on my part
- this thing has trouble with long file names and non alphanumerical characters.
- this thing has trouble with device it incorrectly identiefies or cannot identify. this screws up wonkymedia.
- the communication between the process on the NMH (that serves the flash user interface running on your browser ) and winkymedia is going through a network port itself. There are problems. thye do not do inter process communication but go via network messages... ( this is kind of dumb as it loads the network... )
proof :
21:01:03:625 filescanner thread started
21:01:03:628 LOG_SSDPSDP_notify_packet ### SSDP sending:
NOTIFY * HTTP/1.1
HOST: 239.255.255.250:1900
CACHE-CONTROL: max-age=99999
LOCATION: http://192.168.1.66:9000/DeviceDescription.xml
NT: urn:schemas-upnp-org:service:ContentDirectory:1
NTS: ssdp:alive
SERVER: Linux/2.x.x, UPnP/1.0, pvConnect UPnP SDK/1.0
the filescanner sends messages through network port 1900 to the NMH. Since the filescanner is running on the NMH ... i also have no clue what 239:255:255:250 at 190 is ...
i also see sometimes the following messages fly by
21:01:03:894 [Error] - LOG_HTTP:HTTP_get_header Cannot receive header, clientSocket=13, nBytesReceived=0
21:01:03:894 LOG_HTTP:HTTP_send_receive received no header in HTTP_send_receive, propably client closed socket, URL=http://192.168.1.66:9000/
my suspicion is that , since the various processes running on the NMH ( the indexer , the UI server and all the twonky processes) all intercommunicate through the network port , this is a problem. if the network settings get corrupted ( because false identification , notwork overload or whatever ) the thing jams up.
by cleaning out all the false identieis , letting it identify itself ( important for its own process communication ) and a pc , and then turning off the detection .this solves that problem.
limiting the filesstem to 'clean 32 char max' names solves another problem.
I also see a lot of keep-alive message fly by on the netowrk ( several a second. )
i eventually plugged in my other computers and the roku , let it autodetect for a while and turned this feature back off. Then i plugged in all other devices the NMH has no business with.
So far it still works fine.
I still do see a ton of messages fly by where the NMH is probing the network on existing devices. tey come back with 'device already validated
21:01:04:140 LOG_CLIENT_DB:The entry found in known clients list is already validated (ip=192.168.1.76)
21:01:04:140 LOG_CLIENT_DB:checking http header for entry ip=192.168.1.76, mac=
21:01:04:140 LOG_CLIENT_DB:Checking http header to find a matching client.db entry (ip=192.168.1.76)
21:01:04:141 LOG_CLIENT_DB:Ignoring client with fixed flag = TRUE (ip=192.168.1.76)
( .76 is the pc i am working on right now. )
i don't know why they keep probing. i am not streaming anything and auto detect is turned off ..
anyway i will keep you guys posted on how this evolves.
One thing is for sure. this is another half baked ' broken source' based system.

Answer 1

further observations :
22:24:31:868 LOG_CLIENT_DB:checking http header for entry ip=192.168.1.66, mac=
22:24:31:868 LOG_CLIENT_DB:Checking http header to find a matching client.db entry (ip=192.168.1.66)
22:24:31:869 LOG_CLIENT_DB:HHetting client adaptation to 49 (ip=192.168.1.66)
22:24:36:878 LOG_CLIENT_DB:The entry found in known clients list is already validated (ip=192.168.1.66)
22:24:36:878 LOG_CLIENT_DB:checking http header for entry ip=192.168.1.66, mac=
22:24:36:878 LOG_CLIENT_DB:Checking http header to find a matching client.db entry (ip=192.168.1.66)
22:24:36:879 LOG_CLIENT_DB:HHetting client adaptation to 49 (ip=192.168.1.66)
22:24:37:926 LOG_CLIENT_DB:The entry found in known clients list is already validated (ip=192.168.1.76)
22:24:37:926 LOG_CLIENT_DB:checking http header for entry ip=192.168.1.76, mac=
22:24:37:926 LOG_CLIENT_DB:Checking http header to find a matching client.db entry (ip=192.168.1.76)
22:24:37:927 LOG_CLIENT_DB:Ignoring client with fixed flag = TRUE (ip=192.168.1.76)
22:2
thisnthing keeps on probing itself ... i wonder why. (.66 is the NMH  .76 is my pc ... ) it is also strange it cannot retrieve its own mac address....
Oh , you can turn off the logging features again when done. it only takes time and diskspace on the NMH
And before i get flamed : the comments i make about 'broken source' : i have no gripe with linux. I have a problem with companies that grab a bunch of stuff that is free ,slap it together, sell it for a lot of money and give no support to the people that bought it. They want all the money for no effort .. they turn open source into broken source ...