One minute of data is missing every ten minutes

[peeonsky]

HI, all:

(I can't save the orginal post, the forum always pop a window says "An unexpected error was returned: invalid_data", so I repost the new post, sorry for that)

I am using zabbix to monitor network switches by SNMP, data is collected every 10 seconds, but it is found that 1 minute of data is missing every 10 minutes.
Is there any configuration I can change to fix this ?

OS: CentOS Linux release 8.5.2111
zabbix_server (Zabbix) 7.0.2
Revision d1b0c3308ce 29 July 2024, compilation time: Jul 29 2024 10:37:14

ListenPort=20051
LogFile=/var/log/zabbix/zabbix_server.log
LogFileSize=0
DebugLevel=2
PidFile=/run/zabbix/zabbix_server.pid
SocketDir=/run/zabbix
DBHost=localhost
DBName=zabbix
DBUser=zabbix
DBSocket=/var/lib/mysql/mysql.sock
DBPort=33060
StartPollers=30
StartAgentPollers=5
StartSNMPPollers=10
StartPollersUnreachable=10
StartPingers=10
StartDiscoverers=5
StartHTTPPollers=2
StartVMwareCollectors=10
VMwareCacheSize=48M
SNMPTrapperFile=/var/log/snmptrap/snmptrap.log
CacheSize=256M
HistoryCacheSize=64M
HistoryIndexCacheSize=16M
TrendCacheSize=48M
ValueCacheSize=256M
Timeout=4
LogSlowQueries=3000
StatsAllowedIP=127.0.0.1
StartODBCPollers=30
EnableGlobalScripts=0

[Brambo]

It's unlikely but could it be that there is a discard when unchanged in the pre-processing? (Even if it's there the chance that 6 read outs all return the same data is small) As a test you could remove it.
But maybe the read out of the data at all isn't possible?

[peeonsky]

It seems weired.
There are two situations, the first is that there is indeed no data, and the second is that there is data but it is not drawn。
And, the pre-processing seems nothing to do with this.

[Brambo]

When looking at your graph the first block (most left) it's clear why nothing is drawn as there is nodata, the other to only show the data entries after the gap. So I expect it has the same reason. So your problem is fixing the nodata issue first.
Try increasing your interval time from 10s to e.g. 15s and see if that works better.

[t00m1G]

HI, all:

(I can't save the orginal post, the forum always pop a window says "An unexpected error was returned: invalid_data", so I repost the new post, sorry for that)

I am using zabbix to monitor network switches by SNMP, data is collected every 10 seconds, but it is found that 1 minute of data is missing every 10 minutes.
Is there any configuration I can change to fix this ?

OS: CentOS Linux release 8.5.2111
zabbix_server (Zabbix) 7.0.2
Revision d1b0c3308ce 29 July 2024, compilation time: Jul 29 2024 10:37:14

ListenPort=20051
LogFile=/var/log/zabbix/zabbix_server.log
LogFileSize=0
DebugLevel=2
PidFile=/run/zabbix/zabbix_server.pid
SocketDir=/run/zabbix
DBHost=localhost
DBName=zabbix
DBUser=zabbix
DBSocket=/var/lib/mysql/mysql.sock
DBPort=33060
StartPollers=30
StartAgentPollers=5
StartSNMPPollers=10
StartPollersUnreachable=10
StartPingers=10
StartDiscoverers=5
StartHTTPPollers=2
StartVMwareCollectors=10
VMwareCacheSize=48M
SNMPTrapperFile=/var/log/snmptrap/snmptrap.log
CacheSize=256M
HistoryCacheSize=64M
HistoryIndexCacheSize=16M
TrendCacheSize=48M
ValueCacheSize=256M
Timeout=4
LogSlowQueries=3000
StatsAllowedIP=127.0.0.1
StartODBCPollers=30
EnableGlobalScripts=0

Hey,

You did not mention, do you have that issue with only one device or multiple devices?
This issue might be related to Zabbix but also to your network devices,

How big this environment is, how many devices you monitor via this Zabbix instance? I'm asking because those 10seconds might be the issue when you have a lot of devices which are monitored.

At first, change those 10seconds to default 3m - if everything will be okay, then change those 3m to 1m. You would need to find the right value for your enviroment - those 10second might be not optimal as we can see,
But, at first check default value for this, 3m.

[Markku]

Pay attention to the poller utilizations, maybe they have too much to do (because there are gaps in the value list in your picture). See the graphs provided by the Zabbix server template to see the utilization of various internal components of the server.

And generic advice for forum posting problems: whenever getting an error message from the forum software, open a new tab (or window, doesn't matter) for the forum, complete the Cloudflare check on that tab (you can close this tab/window now), and then retry the operation. Or, even before clicking the Post button, do the same. Unfortunate but that's the deal for now, it seems.

Markku

[Brambo]

Or maybe increases the

StartSNMPPollers=10

to 50 to start with. If I'm not mistaken you could increase it all the way up to 1000 but that will reserve a lot more resources than the default 10
But check your system resources if that doesn't conflict in other processes.