RELEASED: < 1 Minute to 5 Minute Polling Interval Patch

[Hiney]

Ian should have it out by tomorrow night. After that, I will be releasing. Probably on Sunday.

TheWitness

Do i understand that this is in 0.8.6i ?

H

[rony]

Ian should have it out by tomorrow night. After that, I will be releasing. Probably on Sunday.

TheWitness

Do i understand that this is in 0.8.6i ?

H

Yes, it's only for 0.8.6i.

[egarnel]

Quick question: Does this support both 1m & 5m polling? or just one or the other?

[TheWitness]

If you implement this patch, setup a 1 minute cron and then clear your poller cache, you will see that all your 5 minute polls are spread out over each poller cycle.

This patch supports not only 1 minute and 5 minute polling, but 1 hour, 4 hour, 10 hour, 1 day or any other polling interval you create a template for.

So how's that? I do pretty good work sometimes...

TheWitness

[egarnel]

If you implement this patch, setup a 1 minute cron and then clear your poller cache, you will see that all your 5 minute polls are spread out over each poller cycle.

This patch supports not only 1 minute and 5 minute polling, but 1 hour, 4 hour, 10 hour, 1 day or any other polling interval you create a template for.

So how's that? I do pretty good work sometimes...

TheWitness

What do you mean "sometimes...." You guys rock! Cacti is awesome.

[Hiney]

Yes, it's only for 0.8.6i.

I noticed things like "do_hook("poller_top");" and "do_hook("poller_bottom");" being removed in poller.php (other php files have similar things removed) by this patch is that right or is there an error ?

After applying this patch, i have lost the ability to use plugins, so I'm assuming that these things are errors, and just wish to confirm that i need to put them back.

H

[Hiney]

I added "do_hook("config_settings");" back to config_settings.php

and

"global $menu;" and "$plugin_architecture = array(
"version" => '1.0'
);

do_hook("config_arrays");"

back to config_arrays.php, as well as the one mentioned above, and the plugins are back.

I don't know how other are going with this. Surprised no one else has noticed.

H

[Hiney]

Has this been tested on 0.8.6i ???

I noticed the .pdf has v0.8.6f mentioned in it.

I ask as i get no 1 min values, all the graphs that i put on the new regime have "nan" as their value.

anyway, i think i'll have to revert. bummer.

H

PS

the traffic in/out doesn't even appear on the list, in the log ....

• 20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[190] SNMP: v1: 10.1.0.51, dsname: dir_total, oid: .1.3.6.1.4.1.23.2.28.2.14.1.11.7, output: 672256
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[190] SNMP: v1: 10.1.0.51, dsname: dir_used, oid: .1.3.6.1.4.1.23.2.28.2.14.1.12.7, output: 60787
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[185] SNMP: v1: 10.1.0.51, dsname: dir_total, oid: .1.3.6.1.4.1.23.2.28.2.14.1.11.1, output: 2147483647
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[185] SNMP: v1: 10.1.0.51, dsname: dir_used, oid: .1.3.6.1.4.1.23.2.28.2.14.1.12.1, output: 152944
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[180] SNMP: v1: 10.1.0.51, dsname: vol_total, oid: .1.3.6.1.4.1.23.2.28.2.14.1.3.6, output: 6144000
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[180] SNMP: v1: 10.1.0.51, dsname: vol_freeable, oid: .1.3.6.1.4.1.23.2.28.2.14.1.5.6, output: 0
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[180] SNMP: v1: 10.1.0.51, dsname: vol_free, oid: .1.3.6.1.4.1.23.2.28.2.14.1.4.6, output: 5214016
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[170] SNMP: v1: 10.1.0.51, dsname: open_files, oid: .1.3.6.1.4.1.23.2.28.2.7.0, output: 6087
  20061024 09:05:04 - CMDPHP: Poller[0] Host[13] DS[165] SNMP: v1: 10.1.0.51, dsname: fs_reads, oid: .1.3.6.1.4.1.23.2.28.2.1.0, output: 43118698

[cigamit]

<snip>
back to config_arrays.php, as well as the one mentioned above, and the plugins are back.

I don't know how other are going with this. Surprised no one else has noticed.

Not sure what there is to notice, losing those bits of code is exactly what I would expect to happen if you override your already modified Cacti with a different modified code set. The above files were not made for users using the Plugin Architecture, and don't take that into account. If you use the patch I provided, it may work, but no guarantee's. Its not even gauranteed that most plugin's will be happy with the poller running every minute, since I know a few of them hardcode the 5 minute polling interval in, and expect it to work that way. Thold and monitor should work fine, syslog and mactrack also shouldn't have a issue, but I would have to check the rest of them.

[rr]

With the 1minutepolling patch in 0.8.6i and using cactid 0.8.6i,
I am using HP-UX host template to poll CPU Usage via SNMP.
The user,sys,idle and nice cpu times are different data sources for
the same host.. so I get the problem that with 1min polling,
the user cpu time could be measured at a different run than the
idle time etc...

Is this a problem? Should I convert the polling to use one
data source for all four values? How do I do this?

Excerpt of distribution from cacti log:

09:23:01 hostX, hpux_UserCPU,
09:25:01 hostX, hpux_IdleCPU,
09:26:01 hostX, hpux_NiceCPU,
09:27:07 hostX, hpux_SysCPU,
09:28:01 hostX, hpux_UserCPU,
09:30:01 hostX, hpux_IdleCPU,
09:31:01 hostX, hpux_NiceCPU,
09:32:07 hostX, hpux_SysCPU,
09:33:01 hostX, hpux_UserCPU,
09:35:01 hostX, hpux_IdleCPU,
09:36:01 hostX, hpux_NiceCPU,
09:37:07 hostX, hpux_SysCPU,
09:38:01 hostX, hpux_UserCPU,
09:40:01 hostX, hpux_IdleCPU,

[TheWitness]

Yes, you should keep data that you want on one graph within one Data Source. I do not believe that was mentioned in the write-up. Thanks for the usage note.

TheWitness

[rr]

How do I merge many datasources? Preferably as some kind of script...
I have set the poller to 5min again, and adjusted cron. Do I need to revert the patch to be sure this could not happen again?

One suggestion would be to spread out datasources over different runs in such a way that all datasources belonging to one device are processed in the same poller run.

[TheWitness]

You would have to write a script to export the data, merge the XML file using something like a splice command and then import them into a new RRDfile.

TheWitness

[Devil]

Ive found an error in The PDF file attached to this thread.

In section 2.1 "Click Devices..." under section 2 "Create new RRA" should say "Click 'Data Sources' ...
Otherwise a perfect/easy to follow guide...

regards Devil.

[ellerbrockr]

Ive found an error in The PDF file attached to this thread.

In section 2.1 "Click Devices..." under section 2 "Create new RRA" should say "Click 'Data Sources' ...
Otherwise a perfect/easy to follow guide...

regards Devil.

I followed the PDF instructions but somehow when doing step 5 - "Adjust Data Query" after updating the "SNMP - Interface Statistics" with the 1 min poll cacti managed to kill all the "Suggested values" on the page. I did this multiple times and each time cacti managed to kill all the suggested values. Now when I create new graphs they have default values - no good. Either something is 1) broken with cacti 0.8.6.i, 2) I missed something that broke the Data Query or 3) I don