[Mailman-Developers] MAX_RESTARTS

[Ian Eiloart]

Hi,

mailmanctl has a mechanism for restarting failed qrunners. Very handy, but 
I just got bit by MAX_RESTARTS. A resource problem on our mail server meant 
that the virgin runner was failing every half hour or so - so after a few 
hours Mailman stopped sending notifications of various things.

What is the purpose of MAX_RESTARTS? I can see that we might want to stop 
if a runner is restarted ten times a minute, but not ten times a day, 
surely?

Could we get an mm_cfg setting to control this please? That would give an 
easy workaround.

More principled, perhaps, would be to store not the number of restarts, but 
the time of the last restart. If this restart is occuring less than one 
minute after the last, AND MAX_RESTARTS has been exceeded..... or 
something.... then we stop.


-- 
Ian Eiloart
Servers Team
Sussex University ITS