webminstats's Frequently-Asked Questions

FAQ Revised: Tuesday 14 February 2017 11:29:24

Table of Contents

1. config
2. old
3. installation
4. modules
5. usage
6. divers
7. webmin

1. config

1.1. I do not have any link in config page
release < 0.8.6
it seems to be a bug in our language usage, so check your webmin language (in webmin configuration). the english selection seems to always work.
or you can have a direct access to module config with url like http://localhost:10000/sysstats/edit_module.cgi?idx=disk-0.7.5 (for disk module)
fixed in 0.8.6 release

1.2. javascript errors when display graphs
check if your parameter name does not contain 'illegal' characters, such ',' (used by webminstats as separator), ':' (used by rrdtools)

1.3. Is it possible to 'add' a second or third graph ?
Yes it is possible, in the module with the "display" feature. On configuration page, for each parameter, you have a line called "display data", with a default value to "default", others values are "stop", 1 , 2 ... all parameters with the same value will be graphed together, and a link will allow to select the others "display" you can change the display numbers to names with the "display's name" section on top of screen (after "database" and "execution parameter" sections)

1.4. how to change the account which will recieve alarms by mail
the default account is root. You can change it in the module config, but to apply the change, you have to stop/start cron aquisition.

2. old

2.1. disk, irq and network modules don't work
first, check for /bin/bash2 file on your system if it doesn't exist, create a link from /bin/bash to /bin/bash2
second, this modules need to be configured if you have a release before 0.8.3

2.2. error message from cron : $(!tmp) ... bad substitution
it occurs when some bash version 2 syntaxe script is sent to bash version 1.
fixed in webminstat version 0.7.4 (explicit call to bash2 or change to bash 1 syntaxe)

2.3. my network module do not work in 0.8.2 release
this is a bug, it has been fixed in 0.8.3 release. all modules now use "average" consolidation

2.4. I upgrade webminstats, and I loose old data
there is a module upgrade detection, which work when you upgrade a webminstat module. but it does not work on a webminstats global update (because webmin erase all datas). You have to go /etc/webmin/sysstats/modules and copy your rrd files by hand
fixed in 0.8.3 release

2.5. webminstats "freeze"
2 way to search :
  • look the source page (html code) in your browser (alt+u on netscape for exemple)
  • edit /etc/webmin/sysstats/config file and change the debug_mod line to debug_mode=1 then reload the page, and have a look into /var/webmin/sysstats/webminstats.debug file
problem fixed in 0.8.5 release

2.6. installation on suse distribution
to have webminstats work on suse, you have to
  • create a symbolic link from /bin/bash to /bin/bash2 (or change script's headers from bash2 to bash)
  • create a symbolic link from /usr/bin/i386-suse-linux-rrdtool to /usr/bin/rrdtool (or change webminstat's configuration : /etc/webmin/sysstats/config file)
  • create a symbolic link from /usr/bin/lsof to /usr/sbin/lsof (or change filesopen.sh script)
  • create a symbolic link from /var/log/mail to /var/log/maillog (or change mailq.sh script)
all this should not be necessary since release 0.8.9

2.7. Installing module xxxx.tar.gz -> Failed, There is an error in the module file.
webminstat use gnu tools, such tar, with gnu extensions options, such -z for tar this options does not exists on solaris or hpux. there is 2 ways to fix this :
  • install gnu tar
  • modify the systats-lib.pl to add a pipe with gunzip command
fixed in 0.9.2 release

2.8. no data aquisition in 0.9.0
webminstats data aquisition is now lauchned by the script /etc/cron.d/webminstats.cron
on suse, it is not execute (check you /var/log/messages or /var/log/cron)
but if you rename it to /etc/cron.d/webminstats , it seems to work ...
fixed in 0.9.1

3. installation

3.1. does webminstat work on my unix ?
webminstat can work on any unix system, but some acquisition scripts are only written for linux now, see requirement page.
So, want to contribute ?

3.2. no rrd files
during installation, rrdtool was not found, so the data file were not created 2 solutions :
  • create a link from your rrdtool to /usr/bin/rrdtool, remove all webminstats files and reinstall
  • change your /etc/webmin/sysstats/config, suppress your modules and reinstall them. be sure to put the path to the command with the binary

3.3. I install the rpm file and I do not have access to the webminstats
  • you have to go to webmin page (http://localhost:10000)
  • open webmin/webmin users/
  • open root user and add (click) this module in authorized module list

3.4. Module sysstats does not support this operating system
all webmin modules check for compatibility, so compare a list with the installed operating system.
if you are using a linux os not in the list :
  1. send me a mail with the result of the command : grep os_type /etc/webmin/config
  2. you can try to cheat, by replacing the os_type value by "generic-linux" or one the following
    • cobalt-linux
    • corel-linux
    • debian-linux
    • generic-linux
    • gentoo-linux
    • lfs-linux
    • mandrake-linux
    • msc-linux
    • open-linux
    • redhat-linux
    • slackware-linux
    • suse-linux
    • turbo-linux

3.5. Error - Missing Content-Type Header
look at /var/webmin/sysstats/webminstats.debug file for perl messages
  • this can occur with 0.9.x release with a bad rrdtool version (do not match perl version)
  • check if the perl-CGI package is installed

3.6. Can't locate RRDs.pm in @INC
since 0.9.X webminstats use RRDs perl module (part of rrdtool package) to access database. but this module is perl version dependent. for example, I have : $ locate RRDs
which match with my perl 5.6.1
if not, search for a "good" rrdtool version.

3.7. undefined symbol: png_create_write_struct
this occurs with release 0.9.0 on mandrake 8.2 and 9.0
webminstats 0.9 use RRDs perl module instead of rrdtool command to access rrdtool database.
the RRDs.pm is broken in the mandrake packages, so install rrdtool packages from webminstats sourceforge site.

3.8. is webmin software necessary ?
almost yes ... but you can also try without it : see https://sourceforge.net/forum/message.php?msg_id=2468347

3.9. I do not find "wbm" files
"wbm" suffix stands for "webmin", but is not a new format. webmin accepts modules in tar or tar.gz format, and do not care about suffix. So my software is packaged in "tar.gz".

3.10. webminstats is broken after a webmin upgrade
This can occur if you are installing webmin from tar.gz (not from package) , because webminstats use symbolic links to webmin directory
the problem should be fixed in 0.13.0 release

3.11. can not install the rpm package
some distribution (example : mandriva) repackage webmin and change the standard install directories, so wbm-sysstats can not install (can not find /usr/libexec/webmin).
there is ways to solve it :
  • (recommanded) : re-install webmin from official rpm package (http://www.webmin.com/download.html), then install webminstats from rpm
  • keep your webmin, and install webminstats from tar.gz

4. modules

4.1. why do I have only one month of data on screen ?
It's because the files used to store data are too small :) Do not laugh !, we use rrdtool round-robin data file (constant size) to store data, and it was created to store only one month. To store 2 month, use the following commands : cd /etc/webmin/sysstats/modules for fic in `find . -name "*.rrd"`; do rrdtool resize $fic 3 GROW 1000 mv -f resize.rrd $fic done
the release 0.12.0 adds a yearly graph

4.2. the max data info is not the same in hourly/daily/weekly graphs
the first goal is to keep database as little as possible so each one is structured as follow :
  1. a part for "hourly" data, with one record by 60s (1 minute) : 70 records
  2. a part for "daily" data, with one record by 120s (2 minutes): 750 records
  3. a part for "weekly" data, with one record by 900s (15 minute): 650 records
  4. a part for "monthly" data, with one record by 3600s (1 hour) : 2200 records
to pass from hourly to daily (and so on), webminstat ask rrdtool to compute an average on the datas, so you can see (for exemple),
  • on hourly, the load averaged over 1 minute
  • on daily, the load averaged over 2 minutes
  • on weekly, the load averaged over 15 minutes
  • on monthly, the load averaged over 1 hour
the default function is AVERAGE, but you can change the consolidation function for each module : availables functions are MIN, MAX, AVERAGE, LAST

4.3. ERROR: unknown function 'info'
since 0.8.3 release, we use other rrdtool functions as info, dump, restore, which do not exist in old rrdtool release.
update in 1.0.33 or more

4.4. disk graph curves do not match df output
for now, disk capacity is computed from the number of used blocks and the number of blocks.
but the ext2 and ext3 reserve around 5% for root usage in case of file system full, so the real capacity for a lambda a user is 5% lesser than used/total blocks

4.5. filesopen and mysql do not show any graph
in 0.9.2 release, I add "configurable sample rate" and the init is buggy for this 2 modules.
just open the configuration page and change the sample rate to any desired value

4.6. after an upgrade to 0.9.x release, I have no more data
along code rewrite (from shell-script to perl) aquisition script was renamed from sysstats.sh to sysstats.pl, but the cron job can already contain the old name.
to fix, you can :
  • just stop/start data aquisition from main webminstats page
  • edit crontab (/etc/cron.d/webminstats file or "crontab -e" command)

4.7. how to erase all data
before 0.9.5 relase : the easier way is to un-install a module and then re-install it
since 0.9.5, the configuration page has en empty button

4.8. parameters for scripts in custom module
webminstats tests if the given script exists, and disable this parameter if not to avoid mail warnings, so apply the run button, and it will run

4.9. snmp module do not recover informations
first try to get it with the snmpget command
if you are using the perl Net::SNMP module, you have to use numeric oid (not symbolic ones). you may use snmpget -On to convert symbolic oid to numeric ones.

4.10. can I clone a webminstats's module
it is possible possible since release 2.7

5. usage

5.1. how to install/upgrade webminstat
  1. open your browser on webmin main page
  2. go to webmin configuration/webmin module
  3. select install a module, it will replace the old one if exists
  4. since 0.8.3 release, the data and config are keeped

5.2. how to install/upgrade a webminstat module
  1. open your browser on webminstat main page
  2. click on "configure module"
  3. click on "add a new module"

5.3. how to produce batch graphs ?
webminstats usage is interactive, not batch, but you can do it since 0.9.3 release with the batch_graph.pl script :
syntaxe : /usr/libexec/webmin/sysstats/batch_graph.pl module working_directory

5.4. how to debug scripts ?
  • edit your /etc/webmin/sysstats/config and add a line "set -x"
  • then execute the command /etc/webmin/sysstats/sysstats.sh 2>&1 > /tmp/sysstats.output
  • have a look into /tmp/sysstats.output file or send it to me with webmin, webminstats, distribution version
  • don't forget to restore the old config file

5.5. how to remove this cron messages : ERROR: illegal attempt to update using time 1025422862 when last update time ...
this message is sent if data aquisition is launched too nearly
this aquisition is scheduled each minute, as a cron job, so :
  • check your root cron jobs (crontab -l), you should have only one line
    * * * * * /etc/webmin/sysstats/sysstats.sh
  • you use a time synchronisation (rdate, ntpdate ...), which change back your clock
  • cron is not scheduling with a good accuracy, and the jobs are launched with a positive nice, so if the load is very heavy, execution time may be not linear
    in this case, only one (bad) way : change the cron jobs to
    * * * * * /etc/webmin/sysstats/sysstats.sh > /dev/null

5.6. how to remove a module
  1. go to "configure module" page
  2. select the module
  3. click on trash icon (last on top, right)
  4. it will ask you to confirm

5.7. how to stop a module aquisition
  1. go to "configure module" page
  2. select the module
  3. click on pause icon (second on top, right)
  4. the icon change to "record" icon
  5. in the configure module page, you will see in status column "stopped

5.8. I do not see the graphs in "panorama" mode
seems to be a problem with some themes (simple_theme for example), so try a classic one as "old webmin theme".
fixed in 0.9.6

5.9. warnings about too big log files
you can configure the limit in https://localhost:10000/config.cgi?sysstats with the parameter "maximum log size before warning"

5.10. what are 500m values ?
for values below 1, rrdtool change y scale to 1/1000. So 500m stands for 500/1000 (in fact 0.5).

5.11. ERROR while creating graph : Garbage
this occurs with webminstats <= 0.9.8 and rrdtools >= 1.2 because of some incompatiblity between rrdtools 1.0.x and 1.2.x
to solve this, you have to patch the sysstats-lib.pl file by changing lines 962 et 963 with
my $date = strftime "%a %b %e %Hh%Mm%Ss %Y", localtime;
my $sig = "COMMENT:Created $date";
fixed in 0.9.9

5.12. webminstats aquisition is too long
a sensor should not perturb the computer. If the aquisition is too long, the host problem may be cached, so webminstats aquisition script can send a warning.
now the problem is : why is the aquisition script too long ? in general, the problem comes from syslog parsing. In the current release, all the log (mail module, firewall module, internet module...) are parsed on each run, so if the log comes big, it can be very slow. (this will be fixed in a future release).
now the advices :
  • check your log sizes and check the logrotate configuration
  • you can change the time used for warning by going to https://localhost:10000/config.cgi?sysstats and configure "maximum time for aquisition "

5.13. no aquisition with 0.10.4
the installation changed in 0.10.4 : I now use symbolic links instead file copy
on some linux box, it seams to be a problem between cron and the new links :
  • check your cron syslog file
  • check and fix /usr/libexec/webmin/sysstats/webminstats permissions
if you still have problem, you can revert to old system :
rm -f /etc/cron.d/webminstats
cp /usr/libexec/webmin/sysstats/webminstats /etc/cron.d/webminstats

5.14. webminstats error on fedora 14
the problem comes from the rrdtool fedora 14 base package. you should install the fedora update package

5.15. I can not find webminstats in webmin menu
the webminstats module can be found in "system" section with "Historic System Statistics" name.

6. divers

6.1. problem with batch_graph.pl on network module
this is a bug, the network module is not coded as the others ones (relics) fixed in 0.9.4

7. webmin

7.1. can I clone webminstat module using webmin tool
Webminstats was not designed to support cloning (cron file, pos file are the same for the 2 modules). If you still want to use webmin cloning, be sure only one of the 2 modules is running, and use at lease webminstats 2.2.

Copyright (c) 2001 Dan York

This list of questions and answers was generated by makefaq