It came to my knowledge from Amos.
It should restart like apache "restart gracefully"..
I want to know what to do since the sysVinit script is broken and on
CentOS 7 it's handled by systemd anyway..
By the way a nice feature in systemd is that it defines the maximum FD
for the process.
So the question is:
To fix the sysVinit script or fix the restart function?
I do not belive that now I can fix the restart internal function..
If Amos someone will take it to his hands I will push it into the sysVinit.
Eliezer
On 07/09/2014 08:34 PM, Alex Rousskov wrote:
On 07/08/2014 06:34 PM, Eliezer Croitoru wrote:
So, What "squid -krestart" should do?
I am guessing it should issue an "unknown -k action" error and print
usage information because "restart" is not a documented -k action:
" -k
reconfigure|rotate|shutdown|interrupt|kill|debug|check|parse\n"
If we fix the documentation, it should probably nicely stop and then
automatically restart each SMP kid (or worker?) that exited. Perhaps do
nothing if there are no kids (or workers)?
The "automatically restart" part is already handled by the current code,
but, AFAIK, some kids (e.g., diskers) may malfunction if restarted.
We may also decide to remove all support for this option instead. Does
anybody know what the use cases for it are? What does this action try to
preserve compared to a full shutdown followed by regular start?
Disclaimer: I have not tried to research the commit log for this
feature, but if you want to modify/fix it, you should.
Cheers,
Alex.
