[LON-CAPA-admin] httpd: Unit:: messages in /var/log/messages
Raeburn, Stuart
raeburn at msu.edu
Mon Sep 30 14:41:00 EDT 2024
A follow-up on an email thread begun in September 2021 ...
The messages described in this thread were absent with Apache 2.2 (e.g., CentOS 6), but present with Apache 2.4 (e.g., CentOS 7 and later). Starting to find them in the /var/lo/messages was unrelated to the update from LON-CAPA .2.11.2 to 2.11.3.
I filed bug 6976: "Unwanted data in /var/log/messages originating from capa.so with apache 2.4", and made a one line change to capaUnit.c (rev. 1.14) to resolve that bug.
See: bugs.loncapa.org/show_bug.cgi?id=6976 for details.
capaUnit.c rev. 1.14 was included in LON-CAPA 2.11.5 (released August 5th, 2024).
Stuart Raeburn
LON-CAPA Academic Consortium
________________________________________
From: Todd Pfaff <pfaff at rhpcs.mcmaster.ca>
Sent: Sunday, September 12, 2021 7:22 PM
To: Raeburn, Stuart; lon-capa-admin at mail.lon-capa.org
Subject: Re: [LON-CAPA-admin] httpd: Unit:: messages in /var/log/messages
We have been running with these in our loncapa.conf for the past two
years using 2.11.2:
PerlSetVar lonLoadLim 20.00
PerlSetVar lonUserLoadLim 1000
and this is what we're still using with 2.11.3.
> A question I have is: did you see similar entries in /var/log/messages
> when running 2.11.2?
Not that I ever recall noticing.
Todd
On Sun, 12 Sep 2021, Raeburn, Stuart via LON-CAPA-admin wrote:
> Todd,
>
> The login page on your LON-CAPA server at around 5:00 pm ET today (Sept 12) showed:
>
> Server Load: 205.6 percent
> User Load: 10.90 percent
>
> and at 5:10 pm ET today (Sept 12) it showed:
>
> Server Load: 99.5 percent
> User Load: 11.00 percent
>
> The calculation of $loadpercent, i.e., Server Load (as a percentage) is:
>
> my $loadavg;
> {
> my $loadfile=Apache::File->new('/proc/loadavg');
> $loadavg=<$loadfile>;
> }
> $loadavg =~ s/\s.*//g;
>
> my ($loadpercent);
> if ($loadlim) {
> $loadpercent=sprintf("%.1f",100*$loadavg/$loadlim);
> }
>
> where $loadlim is the value for the lonLoadLim in /etc/httpd/conf/loncapa.conf
>
> If your library server is not actually overloaded, then you could either (a) raise the value of lonLoadLim, or (b) set the value of lonLoadLim to 0, in /etc/httpd/conf/loncapa.conf, followed by a reload of the web server.
>
> For a LON-CAPA server which is part of a cluster of LON-CAPA nodes then when LON-CAPA determines a server is at over 100% server load, then after a user authenticates via /adm/login it will switch the user's session to the least loaded available LON-CAPA host listed in the domain configuration for that server.
>
> If your domain runs LON-CAPA as a single (standalone) server then that type of offloading of user sessions from an "overloaded" LON-CAPA server will be unavailable in your LON-CAPA domain.
>
> As regards items such as:
>
> Sep 12 14:23:14 loncapa01 httpd: Unit::[m] = 1 * (1*m^1) (1*s^-2)
>
> in /var/log/messages my guess would be that they originate within capa.so
>
> If I look in the CAPA source code I find
> source.loncapa.org/cgi-bin/cvsweb.cgi/capa/capa51/pProj/capaUnit.c
>
> has this line within print_unit_t(Unit_t *t)
> printf(" Unit::[%s] = %g * ", t->u_symbol, t->u_scale);
>
> and print_unit_t( is called by this line in postorder_utree()
> case U_DERIVED: print_unit_t(node_p);
>
> A question I have is: did you see similar entries in /var/log/messages when running 2.11.2?
>
> The only change I see in the code between 2.11.2 and 2.11.3 for the CAPA engine, which (among other things) performs physical unit checking, is a minor change in capaParserUtils.c to address bug 6900
>
> See:
> source.loncapa.org/cgi-bin/cvsweb.cgi/capa/capa51/pProj/capaParserUtils.c.diff?r1=1.20;r2=1.21
>
> Stuart Raeburn
> LON-CAPA Academic Consortium
> ________________________________________
> From: LON-CAPA-admin <lon-capa-admin-bounces at mail.lon-capa.org> on behalf of Todd Pfaff via LON-CAPA-admin <lon-capa-admin at mail.lon-capa.org>
> Sent: Sunday, September 12, 2021 4:02 PM
> To: lon-capa-admin at mail.lon-capa.org
> Subject: [LON-CAPA-admin] httpd: Unit:: messages in /var/log/messages
>
> We're running LON-CAPA 2.11.3-2021031720 on CentOS 7. We recently
> upgraded from 2.11.2 to 2.11.3 and are starting to see heavy use again
> with the new school term.
>
> We seem to be experiencing some load related problems after the 2.11.3
> upgrade that I don't recall us having last year with 2.11.2. When this is
> happening I am unable to login as Domain Coordinator and monitor Status
> Information.
>
> Other than the 2.11.3 upgrade, and installing all other outstanding CentOS
> 7 updates, nothing has intentionally changed with this LON-CAPA server.
>
> While doing some initial diagnosis of this performance problem, I happened
> to notice that messages like the following are regularly being written to
> /var/log/messages on our LON-CAPA server.
>
> Sep 12 14:23:14 loncapa01 httpd: Unit::[m] = 1 * (1*m^1) (1*s^-2)
> Sep 12 14:23:16 loncapa01 httpd: Unit::[cm] = 0.0001 * (1*m^2)
> Sep 12 14:23:19 loncapa01 httpd: Unit::[deg] = 0.0174533 *
> Sep 12 14:23:19 loncapa01 httpd: Unit::[kg] = 1 * (1*m^-3) (1*kg^1)
>
> I have no idea if this is related to our performance problem, or if it is
> some other insignificant problem, but it is unusual and not something I
> would expect to be finding, neither from httpd nor in /var/log/messages.
> If for no other reason, I'd like to eliminate thes messages since they're
> just log noise consuming disk space.
>
> Any ideas?
>
> Thanks,
> Todd
> _______________________________________________
> LON-CAPA-admin mailing list
> LON-CAPA-admin at mail.lon-capa.org
> http://mail.lon-capa.org/mailman/listinfo/lon-capa-admin
> _______________________________________________
> LON-CAPA-admin mailing list
> LON-CAPA-admin at mail.lon-capa.org
> http://mail.lon-capa.org/mailman/listinfo/lon-capa-admin
>
>
More information about the LON-CAPA-admin
mailing list