Nut upsdev - Dec 2014 - Tiny suggestion for upslog.c

Hi

Thanks for a very nice piece of software.

I have a tiny suggestion for upslog.c.

It adds a signal handler for the USR1 signal, which simply triggers it 
to abort the current sleep and log the requested now.

I find it useful because my upssched-cmd can then killall -USR1 upslog 
to ensure that there is an entry in the log for brief power dropouts 
which would otherwise be unrecorded (because the power is only off for a 
few seconds).

I think that the patch is still valid against the latest source in GIT 
but was originally coded against 2.2.x.


thanks
Andy

--- nut-2.6.5/clients/upslog.c    2014-12-08 23:23:33.274059582 +0000
+++ nut-2.6.5/clients/upslog.c    2014-12-08 23:24:02.906586228 +0000
@@ -76,6 +76,11 @@
      exit_flag = sig;
  }

+static void set_print_now_flag(int sig)
+{
+    /* no need to do anything, the signal will cause sleep to be 
interrupted */
+}
+
  /* handlers: reload on HUP, exit on INT/QUIT/TERM */
  static void setup_signals(void)
  {
@@ -96,6 +101,11 @@
          fatal_with_errno(EXIT_FAILURE, "Can't install SIGQUIT
handler");
      if (sigaction(SIGTERM, &sa, NULL) < 0)
          fatal_with_errno(EXIT_FAILURE, "Can't install SIGTERM
handler");
+
+    sa.sa_handler = set_print_now_flag;
+    if (sigaction(SIGUSR1, &sa, NULL) < 0)
+        fatal_with_errno(EXIT_FAILURE, "Can't install SIGTERM
handler");
+
  }

  static void help(const char *prog)

Hi Andy,

On Dec 8, 2014, at 7:07 PM, Andy Juniper <ajuniper at freeuk.com> wrote:
> It adds a signal handler for the USR1 signal, which simply triggers it to
abort the current sleep and log the requested now.
I will admit that I'm not an expert on how signals interrupt system calls on
all systems, but for what it's worth, the code looks good to me. Is it
correct to assume that you are using this on Linux?
> I find it useful because my upssched-cmd can then killall -USR1 upslog to
ensure that there is an entry in the log for brief power dropouts which would
otherwise be unrecorded (because the power is only off for a few seconds).
Agreed, that sounds very useful.

I don't think we have a standard format for documenting the signals that the
various NUT components respond to. However, I wouldn't want this feature to
remain undocumented. Would you mind writing up a paragraph for the man page, in
docs/man/upslog.txt? We currently describe how the sleep(3) times are
approximate, but this could go in there as well.

Thanks,

-- 
Charles Lepple
clepple at gmail

Hi Charles,

On 10/12/14 14:27, Charles Lepple wrote:
> Hi Andy,
>
> On Dec 8, 2014, at 7:07 PM, Andy Juniper <ajuniper at freeuk.com>
wrote:
>
>> It adds a signal handler for the USR1 signal, which simply triggers it
to abort the current sleep and log the requested now.
> I will admit that I'm not an expert on how signals interrupt system
calls on all systems, but for what it's worth, the code looks good to me. Is
it correct to assume that you are using this on Linux?
Yes, I've been running this on an ancient RedHat 9 system for a few 
years, and I'm just in the process of moving it to EL6.
Certainly on Linux, the signal interrupts the sleep which returns with 
(quoting from the manual page):

    RETURN VALUE
            Zero if the requested time has elapsed, or the number of
    seconds left to sleep, if the call was interrupted by a signal handler.
>> I find it useful because my upssched-cmd can then killall -USR1 upslog
to ensure that there is an entry in the log for brief power dropouts which would
otherwise be unrecorded (because the power is only off for a few seconds).
> Agreed, that sounds very useful.
>
> I don't think we have a standard format for documenting the signals
that the various NUT components respond to. However, I wouldn't want this
feature to remain undocumented. Would you mind writing up a paragraph for the
man page, in docs/man/upslog.txt? We currently describe how the sleep(3) times
are approximate, but this could go in there as well.
Try this - I couldn't see the file you reference but here's a paragraph 
for the manual page - probably fits in after service delays or log rotation.

    ON DEMAND LOGGING
    Sending a USR1 signal to a running 'upslog' process makes it wake
    from the current sleep and log immediately.  This is useful when
    triggered from a upssched event trigger (e.g. AT ONBATT or AT
    ONLINE) to ensure that an entry always exists, even if the power
    goes away for a period of time shorter than that specified by the -i
    argument.

Andy
>
> Thanks,
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.alioth.debian.org/pipermail/nut-upsdev/attachments/20141210/680fdfab/attachment.html>

On 2014-12-09 01:07, Andy Juniper wrote:
> Hi
>
> +    sa.sa_handler = set_print_now_flag;
> +    if (sigaction(SIGUSR1, &sa, NULL) < 0)
> +        fatal_with_errno(EXIT_FAILURE, "Can't install SIGTERM
handler");
> +
>   }
Minor typo there, should be SIGUSR1 =)