Missed snmp-values and wrong graphs

[Voltsifer]

Hello,

I have problems with getting SNMPv2c values from equipements with high network bandwidth.

For example,

border router incoming traffic item (SNMPv2 agent, integer64bit, multiplier 8, Update interval 15 sec, Store value - speed per sec)
Last values:
"2007.Apr.03 19:45:53 57140910
2007.Apr.03 19:45:38 2028452962
2007.Apr.03 19:45:08 42925959
2007.Apr.03 19:44:53 31720155
2007.Apr.03 19:44:38 29567272
2007.Apr.03 19:44:23 2046200865
2007.Apr.03 19:44:08 38764332
2007.Apr.03 19:43:38 98519657
2007.Apr.03 19:42:53 78898011
2007.Apr.03 19:42:23 118986890
2007.Apr.03 19:41:53 102118998
2007.Apr.03 19:41:09 114835215
2007.Apr.03 19:40:53 103782542
2007.Apr.03 19:40:23 89894155"
(Issue ramains in the case of increasing Update interval to 1min - missed values)

Actually this time router's incomming traffic is stable (2.1-2.2 Gb/s) and MRTG-graphs is correct.
There isn't problems at night low-traffic periods.


Graph in attachment: red region in the graph - values from the same device.

What's problem?

Server's CPU is not high loaded...


Zabbix 1.1.7
SunOS 5.10

[Flooter]

Hello!
Any update on the issue? I am having the same problem as well.

[zalex_ua]

Check please description for "Store value" here http://www.zabbix.com/documentation/...fig/items/item
Also see this my picture

[Flooter]

Thank you very much for you help, very much appreciated!


P.S. I wish I could move from 32bit counters to 64bit counters transparently or, at least, be able to manually remove that single spike I have now in history when zabbix picked up new value from new OID (that thousand times bigger then previous one)

Do I have any other options apart from just clearing whole history for the item I moved from 32 to 64bit counters? Like clearing only single value may be?

[zalex_ua]

You can delete the spikes from your database. Check history (or history_uint) and trends (or trends_uint) tables.

*_uint - it's for numeric float items.