Clarification on ModemManager Reconnection Behavior
Dan Williams
dan at ioncontrol.co
Thu Jul 31 12:22:51 UTC 2025
> On Jul 28, 2025, at 3:26 AM, Armand Daussà <adaussa at cactusiot.com> wrote:
>
> Hi Dan,
>
> Thank you for your reply!
>
> In our case, NetworkManager is the service responsible for managing the connections. I’ve checked the configuration of the cellular_connection, which is the entry in NetworkManager for our WWAN interface. Here are the relevant details:
>
> connection.id: cellular_connection
> connection.uuid: dea368fe-e05d-4dad-b8bf-37e67e47fe56
> connection.type: gsm
> connection.autoconnect: yes
> connection.autoconnect-retries: -1 (default)
> connection.auth-retries: -1
> ... (rest omitted for brevity)
>
> My understanding is that the values connection.autoconnect-retries: -1 and connection.auth-retries: -1 should allow for infinite reconnection attempts. Is there any additional configuration we might be missing to ensure that reconnections always occur after transient modem or network issues?
Looking at the nm-settings manpage I see two things. First it says auth-retries is only for 802.1x so that probably doesn’t matter here. But second it says that -1 uses the global default number of retries of 4 which is probably pretty low for you. 0 is supposed to be “forever”; can you try that value instead of -1?
>
> Regarding the modem disconnection messages: we have observed multiple devices experiencing disconnects. However, the <CR><LF>+SIMCARD: NOT AVAILABLE<CR><LF> message appears only on the device included in the logs I shared.
>
> Other devices report different verbose call end reasons such as:
> • [internal] ip-version-mismatch
> • [cm] no-service
> • [internal] mmgdsi-card-event (also on the same device with the SIMCARD error)
> • [cm] emm-detached
> • [internal] pdn-ipv6-call-disallowed
> • [cm] emm-t3417-expired
> • [cm] detach-with-reattach-lte-nw-detach
The ip-version-mismatch and pdn-ipv6-call-disallowed ones may indicate that the APN you’re using isn’t configured for IPv6, but MM is requesting either V6 or V4V6 activation. That can be changed in the NM settings.
>
> From your perspective, do these messages generally point toward modem hardware issues or even physical/network instability, rather than issues with ModemManager itself?
The SIM card errors may point to either a bad SIM or bad SIM slot on that specific modem. The two IP-related ones may be configuration, and the rest look like fairly normal causes that a reconnect should eventually be able to recover from. Note that networks sometimes have backoff timers to prevent reconnect spamming; not sure if that is the case for yours.
>
> Lastly, the modem does not support hot swap, so unfortunately we cannot reinsert the SIM card once removed.
Are you able to power cycle the modem via GPIO or some other method? I’ve had cases where it would be helpful to do that based on a couple factors, like whether the modem hasn’t been registered in a given amount of time even though you know there should be an available network.
Dan
>
> Thanks again for your help!
>
> On Sat, Jul 26, 2025 at 5:29 AM Dan Williams <dan at ioncontrol.co> wrote:
> On Wed, 2025-07-16 at 17:35 +0200, Armand Daussà wrote:
> > Hi,
> >
> > We’ve encountered an issue with modem disconnections and have
> > reviewed the ModemManager logs to better understand the behavior.
> > What we observed is that after a disconnection, ModemManager attempts
> > to reconnect but then shuts down and does not attempt to reconnect
> > again. It only resumes connection attempts after the operating system
> > is rebooted.
> >
> > In situations where the modem experiences issues and cannot establish
> > a connection, ModemManager appears to stop trying entirely until the
> > next system reboot. Eventually, after a reboot, ModemManager is able
> > to successfully initialize the modem and re-establish a connection.
> >
> > Could you please confirm if this is the expected behavior? If not, is
> > there a recommended way to ensure that ModemManager continues trying
> > to reconnect without requiring a system reboot?
>
> ModemManager itself does not actually start/stop the data connection.
>
> That's typically done by a connection manager, like NetworkManager,
> distro network scripts, or custom scripting.
>
> In the NetworkManager case, it will stop asking MM to reconnect after a
> configurable number of failures. It will then start a timer and when
> the timer expires, retry until the configurable number of failures.
> Repeat.
>
> Do you know what connection manager you're using to drive ModemManager?
> If we can figure that out we can nail down the reconnect behavior.
>
> >
> > Is it something fixed in more recent releases?
> >
> > Key timestamps:
> > 16:37:24 -> Modem sends '<CR><LF>+SIMCARD: NOT AVAILABLE<CR><LF>'
> > Five OS Reboots are performed
> > 19:19:21 -> Modem manager starts following last reboot and a
> > successful initialization is completed
> >
> > Logs are attached.
>
> I do notice:
>
> Jul 14 16:37:24 fonts2 ModemManager[498]: <info> [1752503844.286189]
> [modem0/bearer1] bearer call end reason (1): 'generic-unspecified'
> Jul 14 16:37:24 fonts2 ModemManager[498]: <info> [1752503844.286265]
> [modem0/bearer1] bearer verbose call end reason (2,218): [internal]
> mmgdsi-card-event
>
> The call here gets terminated by mmgdsi-card-event which I think is
> *actually* GSDI = Generic Sim Driver Interface card event. Which
> correlates to the +SIMCARD message and indicates that the SIM card or
> the modem's SIM interface decided to hiccup and that terminated the
> call.
>
> I'm not surprised things are broken until a reboot, but if this is a
> physical SIM card you might try removing and re-inserting the SIM and
> see if that makes things work more quickly.
>
> In any case, if you're able to determine what connection manager you
> use, then we can figure out the reconnect behavior.
>
> Dan
>
>
>
> >
> > Thanks in advance!
> >
> > System info:
> > OS:
> > Linux fonts1 6.1.21-v8+ #1642 SMP PREEMPT Mon Apr 3 17:24:16 BST
> > 2023 aarch64
> > > PRETTY_NAME="Debian GNU/Linux 11 (bullseye)"
> > >
> >
> >
> > Hardware
> > > manufacturer: QUALCOMM INCORPORATED
> > > model: 0 (doesn't show but its a SIMCOM S2-107EQ)
> > > firmware revision: LE11B12SIM7600M22
> > > carrier config: ROW_Gen_NonVoLTE
> > > carrier config revision: 05010813
> > >
> > > h/w revision: 10000
> > > supported: gsm-umts, lte
> > > current: gsm-umts, lte
> > >
> >
> >
> > ModemManager version
> > 1.14.12
>
>
>
> --
> Armand Daussà
> RD
> +34 644 552 758
> linkedin.com/company/cactusiot
> cactusiot.com
> Sant Joan de la Salle 42, 08022 Barcelona
> Aquest missatge de correu electrònic i els seus documents adjunts estan dirigits EXCLUSIVAMENT als destinataris especificats. La informació continguda pot ser CONFIDENCIAL i/o estar LEGALMENT PROTEGIDA i no necessàriament reflecteix l'opinió de Cactus Engineering S.L. Si vostè rep aquest missatge per ERROR, si us plau comunicar-ho immediatament al remitent i ELIMINI'L, ja que vostè NO ESTÀ AUTORITZAT al seu ús, revelació, distribució, impressió o còpia de tota o alguna part de la informació continguda. Gràcies.
>
> Este mensaje de correo electrónico y sus documentos adjuntos están dirigidos EXCLUSIVAMENTE a los destinatarios especificados. La información contenida puede ser CONFIDENCIAL y/o estar LEGALMENTE PROTEGIDA y no necesariamente refleja la opinión de Cactus Engineering S.L. Si usted recibe este mensaje por ERROR, por favor comuníqueselo inmediatamente al remitente y ELIMÍNELO ya que usted NO ESTÁ AUTORIZADO al uso, revelación, distribución, impresión o copia de toda o alguna parte de la información contenida. Gracias.
>
> This message and any attached documents are intended ONLY for the use of the individual or entity to which they are addressed. It may contain CONFIDENTIAL and/or LEGALLY PRIVILEGED information and may not necessarily represent the opinion of Cactus Engineering S.L. If you have received this email in error, please notify the sender immediately and DELETE it, as you ARE NOT AUTHORISED to use, disclose, distribute, print, or copy all or any part of the contained information. Thank you.
More information about the ModemManager-devel
mailing list