LTE network shutdown after some time
mlansirinne at forcepoint.com
Wed Jan 16 13:14:17 UTC 2019
Ok, thanks. That clarifies the issue. So I will need to have a daemon polling the status to see when this happens.
Is it so that the mbim kernel module (or mbim-proxy?) don't see this happening and can't forward an event for example in the cdc-wbm0 device file if I select/poll that file? (I mean, I would usually prefer waiting for an event than actively checking the status and have the service down for up to the polling period).
From: Dan Williams <dcbw at redhat.com>
Sent: Tuesday, January 15, 2019 21:14
To: Lansirinne, Mika; libmbim-devel at lists.freedesktop.org
Subject: EXTERNAL: Re: LTE network shutdown after some time
On Tue, 2019-01-15 at 18:28 +0000, Lansirinne, Mika wrote:
> I am testing a couple of LTE modems from Sierra (EM7455 and EM7565
> specifically), and found some strange behavior: after some time of
> working, the LTE network seems to drop the connection without having
> any indication of that happening, other than that the traffic does
> not flow any more of course. The both models work the same way.
This is usually an ISP/provider choice. Many will force-disconnect
devices if they remain connected for a certain period of time, usually
12 - 24 hours. A reconnect (as you've found) is the normal workaround.
Typically the connection manager (be that ModemManager, NetworkManager,
custom scripting, etc) would be used to automatically reconnect when a
ping or some other "health check" fails.
> I am currently very unsure on what level this is happening. To my
> mind the disconnect can perhaps happen at least because the modem
> itself (hardware or firmware level), in mbim library, by the ISP or
> because of my SIM card / data plan. Neither of which make any sense
> since I can't figure out why would one want to disconnect the
> connection (without reconnecting automatically again) if the device
> is in use. And if this really is intended behavior, I wonder if I
> could at least get some event of this happening so that I can act on
> it as a workaround.
> I have connected to the network using the mbimcli commands, with the
> mbim-proxy running on the background:
> # mbimcli -p -d /dev/cdc-wdm0 --query-subscriber-ready-status
> # mbimcli -p -d /dev/cdc-wdm0 --query-registration-state
> # mbimcli -p -d /dev/cdc-wdm0 --attach-packet-service
> # mbimcli -p -d /dev/cdc-wdm0 --register-automatic
> # mbimcli -p -d /dev/cdc-wdm0 --connect=apn=internet
> After that I fetched the ip parameters and set the ip address
> manually since these modems do not support the DHCP:
> # mbimcli -p -d /dev/cdc-wdm0 --query-ip-configuration
> # ip addr...
> # ip route...
> All of that works perfectly.
> I ran a test over a weekend in which I sent one ping packet once in a
> minute, and if that failed then reconnecting the lte. It turned out
> that the connection was dropped exactly every 12 hours and four
> minutes. The query-packet-service-state was showing that "no cells in
> the location area" in that time. Yet the re-connection was
> immediately successful so there really were no actual problems with
> the LTE towers.
> # mbimcli -p -d /dev/cdc-wdm0 --query-packet-service-state
> [/dev/cdc-wdm0] Packet service status:
> Network error: 'no-cells-in-location-area'
> Packet service state: 'detached'
> Available data classes: 'unknown'
> Uplink speed: '0 bps'
> Downlink speed: '0 bps'
> Without sending the ping packets, i.e. just letting the modem idle
> and detecting the outage by the query-packet-service-state command,
> the disconnection may happen much sooner, even less than an hour but
> normally not immediately.
> Would anyone have any ideas if I am just using it wrongly or is there
> some real problem which could be fixed on some level?
> Best regards,
> libmbim-devel mailing list
> libmbim-devel at lists.freedesktop.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the libmbim-devel