Fwd: Re: Wiki advice on running getmail on INBOX access - how does that work?

[c128 mail]

Bit more investigation...

So - the incrontab rule seems to work to the extent that if I directly 
"ls" the cur folder on the system then it's triggered.

However, the only way I seem to be able to be able to get it to trigger 
from a mail client (I use both Thunderbird and Roundcube) is by 
performing some sort of action on the mail that's already in the folder 
e.g. shifting mail in/out, marking mail read, reading an existing mail 
(in Roundcube only, for the last one, which I suspect is related to 
Thunderbird and local caching?).

Either way, I can't seem to get it to trigger on just access to the 
folder its monitoring...

On 18/10/14 21:20, c128 mail wrote:
> Thanks for trying that.
>
> I'm running:
>
> Ubuntu 14.04, Linux 3.4.79 #1 SMP PREEMPT Wed May 14 18:19:18 CST 2014
> armv7l armv7l armv7l GNU/Linux
>
> incrontab 0.5.10
>
> Just found that if I "ls" that directory, the rule is fired.  So - I
> suspect incrontab is fine.
>
> Is this an oddity of Thunderbird access to the particular mail folder, I
> wonder, as that's my use case for testing this?
>
> On 18/10/14 20:58, Gedalya wrote:
>> On 10/18/2014 03:26 PM, c128 mail wrote:
>>> Just re-tried this, and it doesn't seem to fire getmail on access for
>>> me.
>>>
>>> My incrontab is as follows:
>>>
>>> /home/user/Maildir/cur/ IN_ALL_EVENTS,IN_ONESHOT
>>> /home/user/bin/mvmail.sh
>>>
>>> The incrontab rule does work, but only if I make a physical change in
>>> /home/user/Maildir/cur/ e.g. by moving a mail from another folder in
>>> there.  Just accessing the particular inbox doesn't seem to trigger
>>> anything.
>>
>> In my testing I get 3 events every single time I list the test
>> directory, if I remove the IN_ONESHOT.
>> With IN_ONESHOT, I get only one, for the first time.
>> The script on the wiki ends with:
>> incrontab --reload # Rearm the one-shot rule
>> So that makes sense.
>>
>> Debian Jessie, Linux 3.16.5-1 amd64, incron 0.5.10-2
>>
>>>
>>> On 18/10/14 19:54, c128 mail wrote:
>>>> Yeah, I see what you mean - it should trigger by IN_ACCESS (from
>>>> IN_ALL_EVENTS) shouldn't it.  I hadn't previously scanned over the full
>>>> set of events:
>>>> http://manpages.ubuntu.com/manpages/intrepid/man5/incrontab.5.html
>>>>
>>>> I'll set this up again and report back one way or another.
>>>>
>>>> Thanks.
>>>>
>>>> On 18/10/14 19:50, Gedalya wrote:
>>>>> On 10/18/2014 02:38 PM, c128 mail wrote:
>>>>>> > What exactly did happen?
>>>>>>
>>>>>> Oh, yeah, didn't make that clear ;-)
>>>>>>
>>>>>> Nothing happened...not without coaxing.
>>>>>>
>>>>>> If I forced a change to the directory, then it worked - but there
>>>>>> wouldn't be a change to the directory in normal operation, other than
>>>>>> by mail population?
>>>>> It's not supposed to require changes. Note the IN_ALL_EVENTS
>>>>> definition.
>>>>> This would include any attempt to open the directory and take a peek.
>>>>> If you configured it according to the wiki and it's not working
>>>>> then we
>>>>> have troubleshooting to do, but that's the theory.
>>>>> I'm not familiar with incron but I've worked with Linux's inotify.
>>>>>
>>>>>>
>>>>>> On 18/10/14 19:28, Gedalya wrote:
>>>>>>> On 10/18/2014 01:24 PM, c128 mail wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I'm currently running getmail in a cron job every 2 minutes, so I
>>>>>>>> was
>>>>>>>> quite intrigued by this on the wiki:
>>>>>>>>
>>>>>>>> http://wiki2.dovecot.org/HowTo/TriggerGetmailOnIMAPAccess
>>>>>>>>
>>>>>>>> Thing is - I couldn't see how it would work and, when I tried
>>>>>>>> it, it
>>>>>>>> didn't work (at least not for me).
>>>>>>> What exactly did happen?
>>>>>>>
>>>>>>>>
>>>>>>>> It details using incrontab to monitor
>>>>>>>> /home/username/.maildir/cur in
>>>>>>>> order to trigger getmail.
>>>>>>>>
>>>>>>>> However, "cur" won't change unless populated with mail...as an
>>>>>>>> initial
>>>>>>>> result of actually running getmail?  Seems like a chicken and egg
>>>>>>>> situation.
>>>>>>> access != change
>>>>>>>
>>>>>>>>
>>>>>>>> I reckon I'm missing something fundamental :-), but what is it?
>>>>>>>>
>>>>>>>> Ta.
>>>>>>>
>>>>>