Gaps in data for Active Checks

[jasonw]

Why is this happening? What can I do to troubleshoot it? How do I fix it?

Basically, I have an active check set up for a host and I am having "gaps" in the data.

Note that most of the time this value is zero, but I expect to see a solid green line at the bottom, not missing pieces...

I will post the config tomorrow as apparently I have exceeded my attachment space for today...

Also when viewing "Latest Data" for this host, it DOES appear in the list, but is grey (almost like a disabled color), so Zabbix apparently sees some sort of issue with it.

[jasonw]

Here is an example... It might be that no data is coming over... Here is some debug log for another windows counter that I set up:

696:20131104:112715.338 JSON before sending [{
"request":"agent data",
"data":[
{
"host":"MYSERVER",
"key":"perf_counter[\\4\\58]",
"value":"87760896.000000",
"clock":1383593230,
"ns":78979102},
{
"host":"MYSERVER",
"key":"perf_counter[\\238(_Total)\\144]",
"value":"0.000000",
"clock":1383593231,
"ns":83271756},
{
"host":"MYSERVER",
"key":"perf_counter[\\238(0)\\144]",
"value":"4.674581",
"clock":1383593232,
"ns":89427636}],
"clock":1383593235,
"ns":94180126}]
696:20131104:112715.339 JSON back [{
"response":"success",
"info":"Processed 2 Failed 1 Total 3 Seconds spent 0.000076"}]
696:20131104:112715.340 In check_response() response:'{
"response":"success",
"info":"Processed 2 Failed 1 Total 3 Seconds spent 0.000076"}'


Note that I should see this value in a graph somewhere:

"key":"perf_counter[\\238(0)\\144]",
"value":"4.674581",

But, when I look at "Lastest Data" and graph, I do not see anything but zeros, and gaps where there are breaks in the graph. The message above also seems to suggest that there was 1 failure (I guess this counter)...

[jasonw]

"clock" and "ns" is used twice in last record.

I don't think you are correct. One of them is for the metric record, and the other is for the "data" array record. Also, we have almost 1000 hosts on here now. This is only happening for the few counters that I have defined in a custom template for Windows hosts. The Linux ones seem to be doing just fine, but they use the standard templates.

I'm guessing it's a configuration issue, but not sure how to diagnose...

[tchjts1]

What do you have for the "Type of information" for that item? Numeric unsigned or Numeric float?

Try Numeric float if you have it currently set as unsigned, and give it a few minutes.

[jasonw]

It was numeric(unsigned). There are other checks using that, but I changed it to numeric (float) on a few of these to see what happens. I will check it out in a bit!

[tchjts1]

I think that will fix it. We had the same issue as you the other day. Item was set to numeric unsigned, so we had gaps in the data any time the value coming in did not equate to a whole number. Might take like 5 minutes or so until you start seeing steady data.

[jasonw]

Sweet! This indeed did fix some of them (though it took like 30 minutes to update). Now, I am just waiting for the discovery items to catch up... Fingers crossed.

[tchjts1]

Cool. Yeah, discovery stuff takes upwards of an hour.

Just curious, do you have any of your hosts reporting through Zabbix proxy servers? if so, you can adjust how long it takes for them to get new configuration data. By default it is set to 3600 minutes.

[jasonw]

Cool. Yeah, discovery stuff takes upwards of an hour.

Just curious, do you have any of your hosts reporting through Zabbix proxy servers? if so, you can adjust how long it takes for them to get new configuration data. By default it is set to 3600 minutes.

Nice! This fixed everything although this doesn't make logical sense to me as "Numeric (decimal)" should be able to parse "decimals"! Otherwise, it should be called "Integer". Anyway, this works, and I am happy that it does!

Yes, all of our checks go through proxy. We have adjusted that setting in the last hour and will monitor.

[tchjts1]

Yes, all of our checks go through proxy. We have adjusted that setting in the last hour and will monitor.

You'll find you like that setting much better than 3600. Now when you add a new host, it will start sending data in 5 minutes, instead of in an hour. Although, discovery rules stuff still takes up to an hour.

You restarted your proxy process after you made that change right? Cool.