[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: svn commit: r1837599 - /httpd/httpd/branches/2.4.x/STATUS

Hi Yann, I will try to comment inline per patch. Yes, that's always a difficult decision. I think for the "?auto" part it should be easy: it uses a line based key-value format, so adding new keys should be fine for nearly any parser. For the HTML based output the decision is more difficult.

Am 28.08.2018 um 15:54 schrieb Yann Ylavic:

> Those changes look fine (and great) to me, I wanted to +1 but I'm
> wondering if they really belong in 2.4.x since the output of
> mod_status is changed in a way that could break existing parsers.
> I don't know the "policy" here...

On Tue, Aug 7, 2018 at 4:19 PM <rjung@xxxxxxxxxx> wrote:

Author: rjung
Date: Tue Aug  7 14:19:31 2018
New Revision: 1837599

URL: http://svn.apache.org/viewvc?rev=1837599&view=rev
Propose a few monitoring improvements.


Modified: httpd/httpd/branches/2.4.x/STATUS
URL: http://svn.apache.org/viewvc/httpd/httpd/branches/2.4.x/STATUS?rev=1837599&r1=1837598&r2=1837599&view=diff
--- httpd/httpd/branches/2.4.x/STATUS (original)
+++ httpd/httpd/branches/2.4.x/STATUS Tue Aug  7 14:19:31 2018
        2.4.x patch: http://home.apache.org/~jim/patches/socache_redis.patch
        +1: jim,

+   *) mod_proxy: Improve the balancer member data shown
+      in mod_status when "ProxyStatus" is "On":
+      add "busy" count and show byte counts in auto
+      mode always in units of kilobytes.
+      trunk: http://svn.apache.org/r1837588
+      2.4.x patch: svn merge -c 1837588 ^/httpd/httpd/trunk .
+                   (adjust CHANGES)
+      +1: rjung

The first two hunks change the HTML output, the third only the auto output. IMHO the change in the HTML output is so useful, that it the format change should be OK. The added busy value is the most important value to decide, whch of your backends is getting slow.

+   *) mod_status: Complete the data shown for async
+      MPMs in "auto" mode.  Added number of processes,
+      number of stopping processes and number
+      of busy and idle workers.
+      trunk: http://svn.apache.org/r1837589
+      2.4.x patch: svn merge -c 1837589 ^/httpd/httpd/trunk .
+                   (adjust CHANGES)
+      +1: rjung

Only auto changes.

+   *) mod_status: Add cumulated response duration time
+      in milliseconds.
+      trunk: http://svn.apache.org/r1837590
+      2.4.x patch: svn merge -c 1837590 ^/httpd/httpd/trunk .
+                   (adjust CHANGES and include/ap_mmn.h)
+      +1: rjung

auto and HTML changes. The HTML change is for request duration, again a pretty helpful value. I could drop it from the per slot table and only keep it in the top part of the status page as separate new lines.

+   *) mod_status: Cumulate CPU time of exited child
+      processes in the "cu" and "cs" values.
+      Add CPU time of the parent process to the
+      "c" and "s" values.
+      trunk: http://svn.apache.org/r1837595
+      2.4.x patch: svn merge -c 1837595 ^/httpd/httpd/trunk .
+                   (adjust CHANGES and include/ap_mmn.h)
+      +1: rjung

This doesn't change the output format, but fixes the wrong (and mostly useless) CPU values we do provide currently.