Ok…
I think I get the wierd one award for the day…
I have a RB433UAH with a sprint 3G USB (598U) modem connected.
The ppp-client link comes up on boot and gets great thruput… (I have an ext ant and goog signal)
the ppp profile has been tried with “00:00:00”, “23:59:59” and no entry for both the idle time outs and the session time outs.
I am seeing that the session time is counting up normaly when the modem connects, then as the idle time out rises ( a min or so) the session time begins to count DOWN.. when it reaches 00:00:00 the modem disconnects and the ppp-client waits for packets. Upon re-connection, the session time reverts to the value it was at then it began to count down…
I have both NTP client (on this router) and a remote dude monitor (on a remote network) generating traffic to this device, so not sure why the idle is “tripping”, but the count down to disconnect I dont get…
I need an allways on connection via the ppp-client / sprint card..
I have a script to bounce the card at midnight (and yes I have lloked and it is not running and causing the “reset”) to make sure counters dont overflow and to just clean up the connection..
???
Update..
As these are static connections (Static IP) I set the IP address on a ppp-profile (and asigned it ot the ppp-client interface.
Set the idle and session timeouts to 23:59:59
Scripted a bounce every 24Hrs (at 12:00AM)
Now the ppp-client hangs up and re-connects every 1min, 27Sec. Give or take a sec..
There is no init string..
I can’t find ANY information on what the channel lay out (if any) on this card is..
Dial string is ATDT#777
Sent a supout and snap of the logs to support today…
Will advise on any updates..
Craig
I got a responce from support this AM…
- I needed to turn off dial on demand (my DUHHHH) I want it allways on…
Dial on demand will wait for packets before connecting…
- the “hangup” status in my logs is a message from the 598U that the tower has hung up on me…
( HOW RUDE !!!)
I cslled SPRINT, and was wondering (since this is an engineering demo card,) if it was being limited or tuned down.. this was not the case… I was off to support…
After explaining what a MikroTik was… She took some information and deturmaned that I was in range of 5 towers… She then took a look at the array of towers and came up with one that “seemed” to be havin issues… It is the closest to me, but she could not tell if I was on that tower… Off to 2nd level support…
After explaining what a MikroTik is… (there is a theme here…) she too looked at the towers…
She came to the conclusion that I was porobably on the “faulty tower” ( I remember a TV show… anyway..)
So… Sprint belives they have a faulty (RUDE) tower on there hands… and it doesn’t like me…
My task is to take the router home (13 miles) and try it there and see if the towers by the house are more polite…
They have opened a repair ticket on the tower in question…
In short it looks like a sprint problem at this point..
They were impressed with the logging and ability to look at thruput in real time … it made isolation muck easyer for them…
More to come as the saga of “the faulty tower” continues…
Got a call from sprint network support today…
They asked if I could put the card into a windows box…
I did, and it came up and stayed up… HMMMMM…
Signal started out ot 20% then bounced to 60 then 80…
I am going to try another tower today.. (could not last night (forgot))
I wish there was more information available RE: the sprint connection… also I still cant come up with any info on the info/ctrl/data channels for this device !!
Update…
Took the router to the house…
Not good news… Same problem there !!
I am going to try reseting the modem and see if that helps…
If not I will downgrade the router to 3.24 and see if that helps..
Tried resetting th modem (AT&F) then ATZ
No change..
Downgraded to 3.24… no change…
Runs for 2Min 30Sec then dropps the line and re-connects…
Will be pestering support again…
Found some init strings…
Will try them in the AM and update..
Found some information:
Init = ATQ0 V1 E1 S0=0 &C1 &D2 +FCLASS=0
Structure = Data = 0
Command = 1
Info = 2
Same issue… Runs 2 Min 30Sec then drop and re-connect..
Interesting observation: If “TCP MSS” changed from default to YES then run time is 2Min 50Sec.
I am beginning to belive thee is a code issue in the PPP-Client modules…
Spoke with Sprint Tier 2 support…
They are looking for the actual init string used under their windows app.
Next step will have to be having them watch the tower, watch the connection come and go and see if they can see the “reason” for the disconnect if it is at their end or if it is a client side drop.
have you enabled additional logging? I think if you turn on raw, read, write, async, etc you can see more things. Maybe there is something helpful in the logs.
I think you can enable windows to log ppp sessions, and then open up the file and just see what the init string was used. There is some registry key to enable that i believe - search for it and you will find it : )
I scrounged up the INIT string (Will post it later)… NO change…
There are several statements about LCP going up and down…
I will look some more… but it seems to be the LCP going down that takes the link down…
I NEED to get this fixed ASAP !!
I have collected DETAILED logs…
Thanks Sam !!
I will be forwarding them to support…
I am expexting a call from a Sr. Engineer from Sprint and a Sr. Engineer from Sierra Wireless..
If a driver is needed, they will write one and provide it to MikroTik… (If the wish to add it)..
I am also expecting a new USB 598 (replacement card for testing to see if it is just a bad card) and a new 4G card to test compatability…
Both will arrive tommorrow AM…
Will send log files to support tonight and will update as information becomes avail…
Changed out the card…
Up for over an hour now…
4G is not properly seen… (Yet) sent supout to support for update…
If it helps you at all I’ve had good experiences with my own 3G card running on v3.25 with no issues so far.
Device is an E169 Huwaei 3G usb dongle.
Used the following config for it:
allow=chap
command-channel=0
data-channel=0
dial-command=ATDT
info-channel=0
modem-init="AT+CGDCONT=1,\"IP\",\"splns333a1\""
null-modem=no
phone=*99#
port=usb2
user=""
password=""
The modem-init as listed will vary for your provider thou.
So…
It seems to be an issue between the card (USB 598) and the tower.
When the card was replaced, the system came up, then bounced the card once. (about 3 Min)
Then went stable for over 16 Hrs…
I have also sent the SupOut from the 433UAH with a U300 (G4) usb device to Uldis at support to see if the driver can be updated to “see” the 4G/WiMax card..
I have noticed that if the system is idle for an extended period of time that if winbox is opened, that the link will reset.. ( I need to to test the case if traffic is generated first, before winbox is opened if it still resets or not)
I am sending the “offending” card back to Sprint, so they can see what is going on…
For now… If you have the same issue, contact your Sprint repesentitive and arrange for a new card.
PS: I spoke with another user, and they commented that they saw something similar when they quickly cycled the power on a Linsys router with a 598 card in it… that they saw a similar pattern…
So if you power goes down, wait a few before bringing the system back online…
If you enable debug on PPP, you’ll probably see a LCP termination request from the base station. I’m having a similar problem with Alltel and a EVDO USB UM175 (as well as a different radio on Verizon which in my case rules out my USB radio and the carrier).
If you google search “evdo lcp timeout” you’ll get a sampling of the 3000+ matches – here’s what i’ve found…
I followed these instructions for my Asus Eee on Linux for the radio:
http://kenkinder.com/using-verizon-wireless-evdo-pc5740-and-linux/
Note the comment here:
“LCP ECHOs: You’ll notice the lcp-echo-failure and lcp-echo-interval commands. I found that after a minute or so of connectivity, pppd disconnected after some lcp-echos failed. These solved that problem.”
Sure enough, in Linux, this makes a huge difference, without these changes, I disconnect often. With them, it stays up with no drops at all.
I contacted support about this issue with 3.27 on a 433UAH, and he sent me a package with the LCPecho set to every 1 minute (apparently default is 3), which IMPROVED for me, but didn’t stop it all.
I’m waiting to see if the support folks will build a PPP package trying what every discussion online about fixing this talks about:
lcp-echo-failure 4
lcp-echo-interval 65535
I will say the 1 minute echo PPP package build they sent me did get the connection staying up 10-30 minutes (versus the 2.5 minutes the normal one did).
Regards,
Dave
One thing I have noticed…
If you have a connection that is stable over a long time, IE no log entries…
The quickest way to make it drop is to look at the PPP interface/status (winbox), the card will imediatly drop and re-init.
I have also noticed that there seem to be two different “types” of link “fail”..
Onr fails with a “hung up” message and the other contained “hanged up”…
A path for the codders to look at…
I have talked with Franklin the Mfg. the Sprint U300..
The device is a composite device.
Devices:
- USB hub
- Device power management for 3G / 4G power management. (USB only can run one modem at a time (power requirements))
- 3G modem
- 4G WiMax modem
- MicroSD reader.
I have been told that a linux driver is “in the works”…
The 3G is a std qualcom device, the 4G is by beceem..
I havent spoken with them yet…
I also understand that WiMax is similar to 802.xx in that there is firmware to be loaded to the chipset on startup and that there is a portion of the code that runs on the host (similar to a suplicant)..
Still rounding up details..
