[ZBX-15173] Can't monitor traffic on interfaces with speed more then 500 Mbit/s by SNMP Created: 2018 Nov 15 Updated: 2018 Dec 06 Resolved: 2018 Dec 06 |
|
| Status: | Closed |
| Project: | ZABBIX BUGS AND ISSUES |
| Component/s: | None |
| Affects Version/s: | 4.0.1 |
| Fix Version/s: | None |
| Type: | Problem report | Priority: | Minor |
| Reporter: | Ivan Prokudin | Assignee: | Unassigned |
| Resolution: | Won't fix | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Description |
|
Hello! When we are trying to get traffic speed on some interface using SNMP it fails when the speed is much more then 100 MBit/s. I have interfaces with 500 Mbit/s average traffic and I get the situation. Details: Detault SNMP template with Interfaces discovery. And item (Incoming\Outgoing) traffic on interface $1 is checked with 1 minute interval. But values are saved in zabbix once at several minutes or even rarely. I have setup snmpd for test on one of my gigabit servers to show you. Here is it's interface stats exported from zabbix: test на h4: Outgoing traffic on interface virbr0 2018-11-15 18:10:27 1542294627 536632784
And that's all since 17:30.
As I can get the root of the problem is that traffic ammount in snmp is stored in 32bit integer, so it's rotated to zero after 4GB sent. And on gigabit link it can be faster then the minute. So default template settings will not work.
I've resolved the problem by changing 1 minute interval to 30 second, but can't imagine what will I do with 10Gbit/s routers. |
| Comments |
| Comment by Edgars Melveris [ 2018 Nov 16 ] |
|
Hello Ivan! A large integer. The signed range is -9223372036854775808 to 9223372036854775807. The unsigned range is 0 to 18446744073709551615 So the problem should be somewhere else. Can you look at zabbix_server.log and see if there are any errors about this item? Also, take a look at Zabbix server graphs, do you have any problems with Zabbix processes? Please provide screenshot for Zabbix data gathering processes. Please also provide the actual item configuration. |
| Comment by Aigars Kadikis [ 2018 Dec 05 ] |
|
Closing issue due to lack of information. |
| Comment by Ivan Prokudin [ 2018 Dec 05 ] |
|
I'm sorry for long answer. I'm talking about limits in SNMP itself. You can see them in snmpget output: snmpget -v 3 -u authPrivUser -l authPriv -a SHA -x AES -A authpass -X privpass localhost IF-MIB::ifOutOctets.42 See "Counter32"? It's reset on near 4.000.000.000. I understand that it's not zabbix problem directly, but may be you can make some workaround of it cause it will very hard to monitor 10Gbit/s interfaces. |
| Comment by Arturs Lontons [ 2018 Dec 06 ] |
|
Hi, Yes, as you described, the root cause of the issue lies in the 32 bit counter. In the scenario that you have described, you can use the 64 bit counters, which are fully supported by the Zabbix SNMPv2 agent. We are closing this ticket, Since we cannot classify it as a Zabbix bug report
Thank you for your time. |