Logging reduction changes

[Peter Naulls]

On 7/26/21 4:20 AM, Aleksander Morgado wrote:
> Hey,

> 
> My question here would be, how is it possible that this message is
> bothering you so much? It's a debug log that is generated only on
> client allocation. Are you allocating and releasing new clients in a
> loop or something? This should only happen on program startup or when
> operation specific clients are allocated (e.g. a new WDS client during
> a connection attempt).

"Bothering you so much" are your words, not mine.  Most of what I do
is dealing with system logs and dis/reconnets, which can get quite lengthy,
hence these changes.  This is small, but it happens several times so it
can add up if there's a failure or whatever.  Superfluous messages
are also a distraction to those who are less familiar with the system
who might be providing me logs.

> Instead of removing it, I think we could improve it adding more
> information about what type of QMI client is being allocated for
> example.

As I think I mentioned, there's already a following message which does
exactly that.

>>
>> --- a/src/libqmi-glib/qmi-device.c      2021-06-03 12:06:40.442931910 -0400
>> +++ b/src/libqmi-glib/qmi-device.c      2021-06-03 12:07:05.435047942 -0400
>> @@ -346,7 +346,7 @@
>>        /* If the command is not abortable, we'll return the error right away
>>         * to the user. */
>>        if (!__qmi_message_is_abortable (tr->message, tr->message_context)) {
>> -        g_debug ("transaction 0x%x aborted, but message is not abortable",
>> transaction_id);
>> +        //g_debug ("transaction 0x%x aborted, but message is not abortable",
>> transaction_id);
>>            device_release_transaction (self, tr->wait_ctx->key);
>>            transaction_complete_and_free (tr, NULL, abort_error_take);
>>            g_error_free (abort_error_take);
>>
>> I had to suppress this.  In my testing, this would happen every 30 seconds. I
>> don't know if it was some side effect or other code we were running or what.
>> Sure, it's maybe interesting once, but not hundreds of times in a log:
>>
> 
> I've seen this message some times, especially when unplugging a
> device, as the ongoing request gets aborted because the port is gone.
> Not sure why you would see this every 30s; is this maybe something
> related to what you do? I think we should definitely try to understand
> the root cause of the message.

Yes, probably. I don't have a lot of time to do so, but if there's experiments
you want me to run, I can.