LinuxQuestions.org - Novatel Merlin V620 CDMA PC Card: How To?

- Linux - Hardware (https://www.linuxquestions.org/questions/linux-hardware-18/)

- - Novatel Merlin V620 CDMA PC Card: How To? (https://www.linuxquestions.org/questions/linux-hardware-18/novatel-merlin-v620-cdma-pc-card-how-to-300962/)

Looks like we are all working on the same thing. I attempted to follow the directions in the above posts and keep running into the following error.

/usr/sbin/pppd: In file /etc/ppp/peers/verizon: unrecognized option '/dev/ttyUSB0'

I should mention that when I reload the usb modules as discribed the device is recongnized and assigned /dev/ttyUSB0 and /dev/ttyUSB1. dmesg confirms command execution.

Any ideas?

Scott.

On my fedora core 3 build udev doesn't create the devices in /dev/ I manually created them with /sbin/MAKEDEV ttyUSB0 and ttyUSB1 respectively. Seems I can communicate with the modem. Now I just have to get the thing to log in...

Scott.

Fixed Problem with v620 LCP Timeout

After searching around for a bit, I discovered that adding a line to the chat script will cure the LCP Timeout problem for me. My chat script now looks like this:

'' 'AT'
'OK' 'ATQ0V1E0'
'OK' 'ATZ'
'OK' 'AT&F'
'OK' 'ATDT#777'
CONNECT CLIENT

Apparently, Microsoft PPP (RAS server) has some quirk about receiving CLIENT as a string. Apparently, the v620 client-side PPP is emulating that strange "feature."

Hope this helps others.

BTW, I believe that the AT commands other than the ATDT are not useful. In fact, the AT&F is probably undoing anything the commands before it happen to cause.

D. Knisely

Finally got the V620 working in my setup (FreeBSD 5.3) -- sort of. It connects, but I only get low bitrates (~ 40 kbps). I'm in an EVDO area - on my Windows laptop I routinely get 400-700 kbps.

Anyone know if there's some AT commands necessary to connect at EVDO rates? I tried the script just posted by D Knisely and still get low bitrates.

More Testing of v620

I can now connect very reliably, but the PPP session is locking up under any stress loading. The behavior is identical under Fedora Core 3 (with all updates) and Redhat 9 with 2.4.20-8 kernel on two (very) different laptops.

Under heavy load (e.g., large FTP download), the PPP connection remains up, but the Novatel card becomes non-responsive. Outbound packets appear to be transmitted on the PPP link, but no traffic is received. The card keeps blinking green occassionally, but no traffic is received on the PPP link. pppd must be stopped and the card must be removed and reinserted.

I have tried every possible combination of pppd settings with many MRU values, allowing/disallowing every combination of asyncmaps and compression options, with no significant improvement. If I set the TCP/IP rmem (RWIN) down to << 16KBytes, a 9 MB FTP will usually complete, but that is no solution. It strongly appears to be a buffering problem either in the USB tty drivers in linux or in the Novatel card. I have tried to replicate every option that I believe Windows is selecting to no avail.

The transfer rate is very good while it lasts (>40 KBytes/sec). I'm using an external antenna from different locations; I believe the RF conditions are good.

Some observations:

- The settings people are proposing for PAP and CHAP secrets and login information are red herrings; for Verizon Wireless EVDO service and the Novatel card, no authentication is being done to the laptop. You can easily verify this by setting "auth" in the pppd settings. The card refuses to permit authentication of any kind. Actual authentication is performed strictly between the modem and the network. Why Verizon software puts "phonenum@vzw3g.com" and "vzw" in the connectoid in Windows is a mystery; perhaps some devices use that fixed authentication even though it is not useful. The Verizon network uses much stronger authentication between the modem and the network.
- I also believe most of the chat strict AT commands proposed by other users are red herrings. A simple ATDT#777 followed by CONNECT CLIENT works for me. I can't get rid of the CONNECT CLIENT. YMMV. AT&F to reset to defaults might be a good idea.
- The data rate (115200, 230400, etc.) probably has no effect.
- I have not been able to see any differences between crtscts settings, but the pppd "xonxoff" setting seems to make lockup happen earlier than with crtscts.
- Settings for vj header compression surprisingly seem to have little effect either.

I'd be very interested to know if anybody has gotten the Novatel v620 to work stably under load in linux.

D. Knisely

Same problem here (lockup under load). I can replicate the problem consistently and quickly by using the SBC DSL Speed Test web site (just Google that -- it's easy to find).

Using the above information, I have mine connecting, and using Ethereal, I can see an IP address for PPP0. (Running under Fedora Core 2, with the Ricoh PCI-to-PCMCIA adapter, and Kernel version 2.6.58 on a Shuttle X mini-desktop) However, it does not seem to be pulling the DNS information. When in a Winders environment with the Verizon software, it does fine. Is there an additional line that needs to be in the verizon file under /etc/ppp/peers ? At first, it was trying to pull DNS from the onboard NIC settings as ETH0. Once I removed those, I tried hard coding the DNS server info into the Lan config information, but it still isn't resolving DNS. Any suggestions on what I may be missing? Thanks in advance for everyone's posts.

I have the Novatel v620. I have thinkpads. That makes it easy to switch to 'doze if I really really really need to.

But I don't wanna... I have tried Knoppix 3.7 and STD, RH9, SuSE 9.2 and FC3. None have given results other than those noted by dknisely. On 'doze, I get upwards of 70% signal reports and solid connections. That quells the urge to buy a booster antenna. It has stayed on all night. But when we start reviewing what we got, it dies.

I also saved the registry settings associated with ppp, RAS, vzw, and verizon. (Ctrl-F and F3). There really wasn't much help there (hmmmm).

I've been using wvdial. It works correctly if I get an LCP TermReq from verizon. If the link dies, the modem hangs. That requires a reboot; the FC3 usbserial module (kernel v2.6.9-1.667) will not unload. This looks an awful lot like commo issues in the modem... imagine that, ppp and all.

So, rather than bore you with config files and the like... who am I kidding, here are the settings:

Somebody already mentioned that FC3 doesn't come with /dev/ttyUSB0. Here's a script that correctly builds those:

----- ~/buildif -----

Code:

#!/bin/bash

#

#  Written by Bill4

#  Script to load Novatel V620 on Fedora 3

#



if [ ! -c /dev/ttyUSB0 ]; then

        mknod -m 660 /dev/ttyUSB0 c 188 0

        chgrp uucp /dev/ttyUSB0

        echo 'ttyUSB0 created'

fi



if [ ! -c /dev/ttyUSB1 ]; then

        mknod -m 660 /dev/ttyUSB1 c 188 1

        chgrp uucp /dev/ttyUSB1

        echo 'ttyUSB1 created'

fi



  modprobe usbserial vendor=0x1410 product=0x1110

----- end ~/buildif -----

this works as long as the user can create dev files. You can manipulate /etc/groups, right? You need both for the v620. I'd unload the module (modprobe -r usbserial) if FC3 had compiled one that would unload. These are quick, and very dirty... USE AT YOUR OWN RISK! THEY MAY HOSE YOUR SYSTEM, DELETE YOUR FILES, and IMPREGNATE YOUR CAT. get it... good!

Now... I needed a way to tell if the line had gone down. Maybe this will give verizon a hint as well: (do they watch their traffic?)

----- ~/monlink -----

Code:

#!/bin/bash

#

#  Script to pull a small page that larger ones may be pulled

#

#  Written by Bill4

#  15 May 2005

#



cd ~/deletabledirectory



while [ 1 = 1 ]; do

        sleep 20 

        wget -N -U 'Mozilla/5.0.(X11; Linux i686; en-US; rv:1.8a6) Gecko/20050105 Mnenhy/0.7.1' http://www.ietf.org/rfc/rfc1661.txt

        done

----- end ~/monlink -----

After the first one, which didn't load fully the first time, the -N switch compares the header info, and decides not to pull it. This renders a nice small session that stays out if the way of downloading a new kernel (if that ever happens). It does give more action than 'ping -i 5 ns.novell.com'. It also gives you some valuable reading material...

----- /etc/wvdial.conf -----

Code:

Stupid Mode = 1

phone = #777

username = 1234567890@vzw3g.com

password = vzw



[Dialer Defaults]

Modem = /dev/ttyUSB0

Baud = 460800

Init1 = ATZ

Init2 = ATQ0 V1 E1 S0=0 &C1 &D2 +FCLASS=0

ISDN = 0

----- End /etc/wvdial.conf -----

I love that "Stupid Mode" parameter... That basically just starts pppd and lets it handle the connection. Pretty smart if ya ask me. Replace the 1-0 in line 3 with your phone number. The rest of it was built with wvdialconf. Which is what leads me to believe that they are wrong. Nothing worth having is that easy.

WVDIAL(1) still uses /etc/ppp/options, so here's mine:

----- /etc/ppp/options -----

Code:

lock

crtscts

debug

noproxyarp

persist

record /var/log/ppp_messages

noauth

mtu 1500

mru 1500

#lcp-echo-failure 20000

#lcp-echo-interval 4

#lcp-max-configure 10

#lcp-max-failure 10

#lcp-max-terminate 2

kdebug 2

----- End /etc/ppp/options -----

I noticed that the mru seemed to be down around 512 or so, so I force it to 1500. I am also logging my third point of contact off...

SO! here's the kicker:

Quote:

<datestamp> 22:33:54 <hostname> pppd[2638]: Sent 66426 bytes, received 479512 bytes.
<datestamp> 22:33:54 <hostname> pppd[2638]: Exit.
<datestamp> 22:34:05 <hostname> pppd[2848]: pppd 2.4.2 started by root, uid 0
<datestamp> 22:34:06 <hostname> chat[2850]: send (AT^M)
<datestamp> 22:34:06 <hostname> chat[2850]: expect (OK)
<datestamp> 22:34:24 <hostname> chat[2850]: ~!E^@^B(}]^T^@^@,^F"DA%[yFU ^E^@P^A^Hv},^FU4k)P^P@^@(^\^@^@dbTxZKqv3LfEGRs+b0xSS%2fCmNRvZf7GD6YI
<datestamp> 22:34:24 <hostname> chat[2850]: %2fcTw%3d%3d"><IMG SRC="/images/nextarrow.gif" WIDTH=23 HEIGHT=23 BORDER=0 ALT="
<datestamp> 22:34:24 <hostname> chat[2850]: Next"></A></TD>
<datestamp> 22:34:24 <hostname> chat[2850]: </TR>
<datestamp> 22:34:24 <hostname> chat[2850]: </TABLE>
<datestamp> 22:34:24 <hostname> chat[2850]: </DIV>
<datestamp> 22:34:24 <hostname> chat[2850]: <P>
<datestamp> 22:34:24 <hostname> chat[2850]:
<datestamp> 22:34:24 <hostname> last message repeated 10 times
<datestamp> 22:34:24 <hostname> chat[2850]: </TD>
<datestamp> 22:34:24 <hostname> chat[2850]: </TR>
<datestamp> 22:34:24 <hostname> chat[2850]: </TABLE>
<datestamp> 22:34:24 <hostname> chat[2850]: 
<datestamp> 22:34:24 <hostname> chat[2850]:
<datestamp> 22:34:24 <hostname> chat[2850]: <BR><IMG SRC="/common/in_images/pixel.gif" WIDTH="1" HEIGHT="30" ALT="" BORDER=
<datestamp> 22:34:24 <hostname> chat[2850]: "0" ><BR>
<datestamp> 22:34:24 <hostname> chat[2850]:
<datestamp> 22:34:24 <hostname> chat[2850]: <TABLE WIDTH="549" CELLPADDING="0" CELLSPACING="0" BORDER="0">
<datestamp> 22:34:24 <hostname> chat[2850]: <TR>
<datestamp> 22:34:24 <hostname> chat[2850]: <TD ALIGN="CENTER">
<datestamp> 22:34:24 <hostname> chat[2850]:
<datestamp> 22:34:24 <hostname> chat[2850]: 
<datestamp> 22:34:24 <hostname> chat[2850]:
<datestamp> 22:34:51 <hostname> last message repeated 9 times
<datestamp> 22:34:51 <hostname> chat[2850]: alarm
<datestamp> 22:34:51 <hostname> chat[2850]: Failed
<datestamp> 22:34:51 <hostname> pppd[2848]: Connect script failed

For the n00bz, that is html in my chat logs. The modem is still active, and passing packets after pppd ends "gracefully".

So... The modem is dying, but the sessions are still running, time to find an AT command reference. There is a list of "acceptable" AT commands in this thread. wvdial(1) is differing, not me. I fire the modem up again and find these:

http://www.cellular.co.za/hayesat.htm (ads, popups)
http://readthetruth.com/modems.htm (Clean enuf for wget) and
http://www.totse.com/en/technology/t...ons/hayes.html (Radical Open source site)

You choose your own poison.

And I dump wvdial... Now I open the link with this:

Code:

ppp call verizon

which requires some instructions in /etc/ppp/peers/verizon. So there:

Code:

hide-password

/dev/ttyUSB0

460800

connect "/usr/sbin/chat -v -f /etc/chatscripts/verizon"

defaultroute

user "1234567890@vzw3g.com"

remotename verizon

ipparam verizon

usepeerdns

which calls for /etc/chatscripts/verizon (the focus of this post):

Code:

TIMEOUT 15

'' 'AT'

'OK' 'ATQ0V1E0'

'OK' 'ATZ'

'OK' 'AT&F'

'OK' 'ATDT#777'

CONNECT CLIENT

But I still get html in my logs if I invoke pppd after the link has died. The timeout is so short because it can be (another reason not to get the antenna). AT all by itself just returns OK.
The next line issues some changes:
Q0 - STATUS DISPLAY ON
V1 - STATUS CODES SENT
E0 - DISABLE LOCAL CHARACTER ECHO
and downward:
ATZ - RESET TO POWER UP CONDITION (CLEARS ALL REGISTERS)
AT&F - RESTORE TO FACTORY CONFIGURATION
ATDT - Dial using tones
CONNECT CLIENT (OK sure, it works as good as not, but changes nothing)

So... starting from the lowest layer, I'm adding &V to a new line.

Code:

 chat[2842]: ^M 

 chat[2842]: &C: 2; &D: 2; &F: 0; E: 1; L: 0; M: 0; Q: 0; V: 1; X: 4; Z: 0; S0: 0;^M 

 chat[2842]: S3: 13; S4: 10; S5: 8; S6: 2; S7: 50; S8: 2; S9: 6; S10: 14; S11: 95;^M 

 chat[2842]: +FCLASS: 0; +ICF: 3,3; +IFC: 2,2; +IPR: 115200; +DR: 0; +DS: 0,0,2048,6;^M 

 chat[2842]: +CDR: 0; +CDS: 0,1,2048,6; +CFC: 0; +CFG: ""; +CMUX: C,2; +CQD: 10;^M  

 chat[2842]: +CRC: 0; +CRM: 2; +CTA: 0; +CXT: 0; +EB: 1,0,30; +EFCS: 1; +ER: 0;^M 

 chat[2842]: +ES: 3,0,2; +ESR: 1; +ETBM: 1,1,20; +ILRR: 0; +MA: ; +MR: 0; +MS: ;^M  

 chat[2842]: +MV18R: 0; +MV18S: 0,0,0; +FAA: 0; +FAP: 0,0,0; +FBO: 0; +FBU: 0;^M 

 chat[2842]: +FCQ: 1,0; +FCC: 0,1,0,0,0,0,0,0;  +FCR: 0; +FCT: 1E; +FEA: 0;^M 

 chat[2842]: +FFC: 0,0,0,0; +FHS: 0; +FIE: 0; +FIP: 0; +FIS: 0,1,0,0,0,0,0,0;^M 

 chat[2842]: +FLI: ""; +FLO: 1; +FLP: 0; +FMS: 0; +FNR: 0,0,0,0; +FNS: ""; +FPA: "";^M 

 chat[2842]: +FPI: ""; +FPP: 0; +FPR: 8; +FPS: 1; +FPW: ""; +FRQ: 0,0; +FRY: 0;^M 

 chat[2842]: +FSA: ""; +FSP: 0^M 

 chat[2842]: ^M 

 chat[2842]: OK

and that is all for now...

OH! and I forwarded my syslogd to a listening server... just for kicks...

Keeping on keeping on...

B.

OK folks...

Here is a 195 page reference to the codes I posted last nite...

http://www.multitech.com/DOCUMENTS/C...s/S000294D.pdf

Is there anybody.... OUT THERE???

I just got the v620 working on my Thinkpad T30 with Ubuntu 5.04 (debian).

Here's what I did:

1.) Use the script posted above to create the ports:

Code:

#!/bin/bash

#

#  Written by Bill4

#  Script to load Novatel V620 on Fedora 3

#



if [ ! -c /dev/ttyUSB0 ]; then

        mknod -m 660 /dev/ttyUSB0 c 188 0

        chgrp uucp /dev/ttyUSB0

        echo 'ttyUSB0 created'

fi



if [ ! -c /dev/ttyUSB1 ]; then

        mknod -m 660 /dev/ttyUSB1 c 188 1

        chgrp uucp /dev/ttyUSB1

        echo 'ttyUSB1 created'

fi



  modprobe usbserial vendor=0x1410 product=0x1110

2.) vi /etc/ppp/peers/verizon

Code:

connect '/usr/sbin/chat -v -t3 -f /etc/ppp/verizon-chat'

noauth

debug

/dev/ttyUSB0

230400

defaultroute

usepeerdns

ipparam rtt

3.) vi /etc/ppp/verizon-chat

Code:

'' 'AT'

'OK' 'ATQ0V1E0'

'OK' 'ATZ'

'OK' 'AT&F'

'OK' 'ATDT#777'

CONNECT CLIENT

4.) vi /etc/ppp/pap-secrets (add the folowing line)

Code:

"123456789@myvzw.com" * "vzw" *

*change the number

5.) vi /etc/ppp/options

Code:

comment out all the lcp options

6.) pppd call verizon

abhopper, et al,

K00l! Glad I could help! Just a few questions.

Does it stay up while you check a map on mapquest?

Does it stay up while you bring up securityfocus.com?

If so, can you post your kernel config?

The procedure I outlined above does not work for me. It disconnects me every 2.5 minutes. I haven't had a time to experiment more.

Here's the log after connecting to Broadband access on Windows:

Code:

05-22-2005 09:22:23.780 - File: C:\WINDOWS\system32\tapisrv.dll, Version 5.1.2600  

05-22-2005 09:22:23.780 - File: C:\WINDOWS\system32\unimdm.tsp, Version 5.1.2600  

05-22-2005 09:22:23.780 - File: C:\WINDOWS\system32\unimdmat.dll, Version 5.1.2600  

05-22-2005 09:22:23.780 - File: C:\WINDOWS\system32\uniplat.dll, Version 5.1.2600  

05-22-2005 09:22:23.780 - File: C:\WINDOWS\system32\drivers\modem.sys, Version 5.1.2600  

05-22-2005 09:22:23.780 - File: C:\WINDOWS\system32\modemui.dll, Version 5.1.2600  

05-22-2005 09:22:23.780 - File: C:\WINDOWS\system32\mdminst.dll, Version 5.1.2600  

05-22-2005 09:22:23.780 - Modem type: Novatel Wireless Merlin CDMA EV-DO Modem

05-22-2005 09:22:23.780 - Modem inf path: oem1.inf

05-22-2005 09:22:23.780 - Modem inf section: Modem0

05-22-2005 09:22:23.780 - Matching hardware ID: usb\vid_1410&pid_1110&mi_00

05-22-2005 09:22:23.800 - 230400,8,N,1, ctsfl=1, rtsctl=2

05-22-2005 09:22:23.810 - Initializing modem.

05-22-2005 09:22:23.810 - CD is high while initializing modem.

05-22-2005 09:22:23.820 - Send: AT<cr>

05-22-2005 09:22:23.820 - Recv: AT<cr>

05-22-2005 09:22:23.820 - Command Echo

05-22-2005 09:22:23.820 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:22:23.820 - Interpreted response: OK

05-22-2005 09:22:23.830 - Send: AT&F&D2&C1E0V1S0=0<cr>

05-22-2005 09:22:23.840 - Recv: AT&F&D2&C1E0V1S0=0<cr>

05-22-2005 09:22:23.840 - Command Echo

05-22-2005 09:22:23.871 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:22:23.871 - Interpreted response: OK

05-22-2005 09:22:23.881 - Send: AT+IFC=2,2<cr>

05-22-2005 09:22:23.881 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:22:23.881 - Interpreted response: OK

05-22-2005 09:22:23.881 - Waiting for a call.

05-22-2005 09:22:23.891 - Send: ATS0=0<cr>

05-22-2005 09:22:23.891 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:22:23.891 - Interpreted response: OK

05-22-2005 09:22:23.891 - 230400,8,N,1, ctsfl=1, rtsctl=2

05-22-2005 09:22:23.901 - Initializing modem.

05-22-2005 09:22:23.911 - Send: AT<cr>

05-22-2005 09:22:23.921 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:22:23.921 - Interpreted response: OK

05-22-2005 09:22:23.931 - Send: AT&F&D2&C1E0V1S0=0<cr>

05-22-2005 09:22:23.961 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:22:23.961 - Interpreted response: OK

05-22-2005 09:22:23.971 - Send: AT+IFC=2,2<cr>

05-22-2005 09:22:23.981 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:22:23.981 - Interpreted response: OK

05-22-2005 09:22:23.981 - Dialing.

05-22-2005 09:22:23.991 - Send: ATD####<cr>

05-22-2005 09:22:30.230 - Recv: <cr><lf>CONNECT<cr><lf>

05-22-2005 09:22:30.230 - Interpreted response: Connect

05-22-2005 09:22:30.230 - Receive Connect but CD was low, Waiting for signal to go high

05-22-2005 09:22:30.250 - CD has been raised

05-22-2005 09:22:30.250 - Connection established at 230400bps.

05-22-2005 09:22:30.250 - Error-control off or unknown.

05-22-2005 09:22:30.250 - Data compression off or unknown.

05-22-2005 09:23:00.253 - Read: Total: 18654, Per/Sec: 620, Written: Total: 4035, Per/Sec: 133

05-22-2005 09:25:00.255 - Read: Total: 325217, Per/Sec: 2554, Written: Total: 84915, Per/Sec: 674

05-22-2005 09:26:18.608 - Hanging up the modem.

05-22-2005 09:26:18.608 - Hardware hangup by lowering DTR.

05-22-2005 09:26:19.179 - Detected CD dropped from lowering DTR

05-22-2005 09:26:19.179 - Recv: <cr><lf>NO CARRIER<cr><lf>

05-22-2005 09:26:19.179 - Interpreted response: No Carrier

05-22-2005 09:26:19.189 - Send: ATH E1<cr>

05-22-2005 09:26:19.219 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:26:19.219 - Interpreted response: OK

05-22-2005 09:26:19.219 - 230400,8,N,1, ctsfl=1, rtsctl=2

05-22-2005 09:26:19.229 - Initializing modem.

05-22-2005 09:26:19.239 - Send: AT<cr>

05-22-2005 09:26:19.239 - Recv: AT<cr>

05-22-2005 09:26:19.239 - Command Echo

05-22-2005 09:26:19.239 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:26:19.239 - Interpreted response: OK

05-22-2005 09:26:19.249 - Send: AT&F&D2&C1E0V1S0=0<cr>

05-22-2005 09:26:19.259 - Recv: AT&F&D2&C1E0V1S0=0<cr>

05-22-2005 09:26:19.259 - Command Echo

05-22-2005 09:26:19.299 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:26:19.299 - Interpreted response: OK

05-22-2005 09:26:19.309 - Send: AT+IFC=2,2<cr>

05-22-2005 09:26:19.309 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:26:19.309 - Interpreted response: OK

05-22-2005 09:26:19.309 - Waiting for a call.

05-22-2005 09:26:19.319 - Send: ATS0=0<cr>

05-22-2005 09:26:19.319 - Recv: <cr><lf>OK<cr><lf>

05-22-2005 09:26:19.319 - Interpreted response: OK

05-22-2005 09:26:19.329 - Session Statistics:

05-22-2005 09:26:19.329 -                Reads : 46 bytes

05-22-2005 09:26:19.329 -                Writes: 40 bytes

I cannot get a reliable connection on my system. It either dies under load when I try to download a file, or it disconnects are 2.5 minutes. Has anyone found a solution to these issues?

Verizon V620 in VMware 5.0 Win XP Pro SP2 & Win 2000 Pro

A different approach of using the Verizon V620 to connect Linux to the Web is using the V620 in a virtual machine created with VMware 5.0. Ok, ok, the guest OS is Win XP Pro SP 2 or Win 2000 Pro but 'just' used as an internet gateway for the time...

>> Ingredients I used for this:

(1) IBM Thinkpad T22, 512 MB, 100GB
(2) Ubuntu 5.04 Hoary Hedgehog
(3) V620
(4)VMware 5.0
(5) Win XP Pro SP 2 (Win 2000 Pro works as well)

<<

>> The recipe:

The Virtual Win XP Guest can have a minimal flexible hard disk (1 GB) and minimum RAM (128 for1 Win XP SP2 and 64 for Win 2000). The RAM can be set to the minimum because the machine will be just the slave for internet connectivity running only the VZAccess Manager and two network connections.

The guest network settings should be set to 'Bridged connected directly to the physical network'.

Install Win XP SP2 without the V620 plugged into the PCMCIA slot.

Do NOT load 'modprobe usbserial vendor=0x1410 product=0x1110' (see explanation further below). If you have loaded it, unload it with 'modprobe -r usbserial'

Once the guest OS has been installed, start the VM.

Install the VZAccess Manager from the Verizon CD and follow the steps until being asked to insert the V620 to be recognized. Plug the V620 in the PCMCIA slot. The VMware host shows after a little while (< 1 min) in the lower right corner of the main VMware window the a blue USB icon, which means that the V620 has been recognized as a USB device by the VM. Don't worry about the USB and VZAccess Manager looking for PCMCIA card. As soon as the USB icon shows up the guest OS (i.e. Win XP Pro SP 2 or Win 2000 Pro) will tell that new hardware was found and will start installing the device drivers. The VM behaves exactly the same as a real Win XP systems. When asked by the VZAccess Manager where to look for the V620 DON'T select USB or Serial but PCMCIA as usual.
The VZAccess Manager will complete the installation as usual followed by the Venturi Client install.
After that the VM will reboot. If after the reboot the V620 is not properly recognized by the system, unplug it briefly and reinsert it. That will do the recognition as a USB device by VMware.
Check the internet connection by loading MS Explorer. The VM should have internet access in the usual fast V620 speed and the card/modem won't time out!

The VM has now essentially two network connections. One via the V620 to the internet and one for to connect the VM to the Linux host. I prefer to have the little network icons visible on the status bar of the VM to follow the network activity (I know it's just blinking blue lights ;-)

The next step is rather easy: Within the VM go into the Windows network settings (should be called by default 'NationalAccess - BroadbandAccess Properties') and enable under sharing the 'Internet Connection Sharing for this connection') and select the ports in the 'Settings' of 'Sharing' to support HTTP 80 etc.

That's pretty much it. Your Linux host is now ready to automatically connect via 'lo' to the shared internet connection of the Windows guest OS in the VM.

To stop the session, first disconnect the card. Then power it of with the VZAccess Manager. The power off is helpful (necessary?) to avoid the Linux host later to hang at the shutdown while trying to power off the PCMCIA device (which is controlled by the VM which is no longer in existence at shutdown and left the V620 uncontrollable if not powered off). One could also just unplug the V620 but again that may cause the Linux host some hick-ups at shutdown.

This recipe should work stable and (!!) doesn't suck too many resources: The Linux host CPU goes up to a max of 9% during heavy network traffic in and out of the VM and the rest of the VM is idle all the time, window minimized - no action (don't enable 'cool' screen saver in the VM - the VM is just the internet Hub...).
<<

>> Addendum:
!!
The resources of the V620 have to be free (i.e. not managed by the Linux Host. That means usbserial should not be loaded with the V620 parameters (i.e. DO NOT load usbserial with 'modprobe usbserial vendor=0x1410 product=0x1110' because that gives the Linux OS the control over the device as a serial USB device as described earlier in this threat). My device manager showed the 'Novatel Wireless Merlin CDMA' with two branches listed each as 'Vendor Specific Interface'.
!!

The Linux web browser settings have to be set to 'Connected directly to the internet'. Proxy or so doesn't work.

I read a lot about the poor performance of the VMware VMs under Linux but I have to say that I am excited about it. Good performance, great network flexibility and device recognition with an excellent graphic makes me actually considering to buy the latest version 5.0 (I run it as a 30 days trial). The Snapshot manager is just fantastic.

The recipe won't work with Win4Lin because of the USB device required within the VM. I had no luck installing Win4Lin with accelerator anyway. Argh.
<<

Hope this works as good for you as it does for me (writing this right now under the described connection which has been up for the last 1:27 hrs without breaking/timing out).

Cheers
------------------------------------------------------------------------------------------------------------------------------------------------------------------

Just repeating what I said in a previous post, but I had to add these lines to my pppd options in order to stay connected longer than a minute:

novj
novjccomp

--Ian