Craig, as you suspect the issue originating with the MQTT keepalive config or the clock event optimization, did you test those? As a theory: if the clock events were the only transmissions keeping your MQTT connection from dropping while parked, the effect could be due to the retransmission of all metrics on reconnect. You could easily test that by re-enabling the clock events, or by sending a custom event every minute from a script. You could also try a smaller keepalive value (e.g. a minute?) or determine and apply the keepalive interval of your MQTT hub. Regards, Michael Am 24.07.25 um 20:04 schrieb Craig Leres:
On 7/13/25 14:23, Michael Balzer wrote:
I know of no other reports regarding increased traffic. But there are few using MQTT, who also need to monitor their data volume, so maybe you're just the only one noticing.
Yet I still think testing in chronological order is not going to produce sensible results, as you mix unfinished intermediate commits from different development branchs, their commit order normally does not match the chronological order of each other or of the master branch. As written before, you need to concentrate on single branches when doing that. The recommended tool for narrowing down bugs by commits is git bisect.
In your case, 78a49a59 is not a master commit, it's one part of pull request #1081. Just as an example, commit 98606bb5304a5c2b832ffbb0d26391a632e5fc17 is dated after 78a49a59 but was added to master before. So it doesn't make any sense to test 78a49a59, unless you're on the #1081 development branch. Use gitk, git log -- graph, or similar to view the branch/commit associations.
Ok, I learned how to use git bisect. I think 47097936 is the last "good" version and 0be0490e is the first "bad" version:
sea 4 % git bisect bad 0be0490e8b3a7224aaee8190dd8dc42bb4a71540 is the first bad commit commit 0be0490e8b3a7224aaee8190dd8dc42bb4a71540 (HEAD) Author: chris8086 <chris-github@ee.eclipse1.net> Date: Sat Nov 16 07:23:10 2024 +0000
Add configurability of MQTT keepalive timer, and prevent clock events being published every minute
vehicle/OVMS.V3/components/ovms_server_v3/src/ovms_server_v3.cpp | 4 ++++ vehicle/OVMS.V3/components/ovms_server_v3/src/ovms_server_v3.h | 1 + 2 files changed, 5 insertions(+)
But there the diff between these two hashes have a LOT of changes so I'm not sure how to proceed.
I also interacted with hologram support (I had a couple of days while running a "bad" version with zero traffic seen by any sim!). The csr volunteered:
I also wanted to point out there are some zero byte sessions occurring. This can be common, but if not, please review this information.
https://support.hologram.io/hc/en-us/articles/360040668834-Find-and-troubles...
I attached a screen grab for the "recent sessions" tab for the sim on the module running 0be0490e. Not sure if this is interesting (but this is another way to see lots of cellular data for a module that has been parked and on wifi for almost a month).
Suggestions for my next move? I'm still not understanding why there would be so much cellular traffic for a module that is parked and on wifi for many days in a row.
I also included my git bisect log.
Craig
sea 9 % git bisect log git bisect start # status: waiting for both good and bad commits # bad: [78a49a596792e2da5338d5c4f1d719490a674164] Add ability to disable legacy topic. Add new configs to web UI and changes.txt. git bisect bad 78a49a596792e2da5338d5c4f1d719490a674164 # status: waiting for good commit(s), bad commit known # good: [bc1f75c950ae7d85b62bff589e1571ba83df3e62] Merge branch 'openvehicles:master' into SmartEQ git bisect good bc1f75c950ae7d85b62bff589e1571ba83df3e62 # good: [470979369be56fe18eae0df90f1619afdd27e850] Manual: fix syntax & spelling errors git bisect good 470979369be56fe18eae0df90f1619afdd27e850 # bad: [aa6c73978bf43ebd16ba396a989b30a7f6fbece1] Merge branch 'openvehicles:master' into optimise-mqtt git bisect bad aa6c73978bf43ebd16ba396a989b30a7f6fbece1 # bad: [0be0490e8b3a7224aaee8190dd8dc42bb4a71540] Add configurability of MQTT keepalive timer, and prevent clock events being published every minute git bisect bad 0be0490e8b3a7224aaee8190dd8dc42bb4a71540 # first bad commit: [0be0490e8b3a7224aaee8190dd8dc42bb4a71540] Add configurability of MQTT keepalive timer, and prevent clock events being published every minute
-- Michael Balzer * Am Rahmen 5 * D-58313 Herdecke Fon 02330 9104094 * Handy 0176 20698926