-
Incident report
-
Resolution: Duplicate
-
Critical
-
None
-
2.2.5
-
None
Probably its somehow connected with ZBX-8528. Algorithm how SNMP poller decides how many OIDs to get per request seems to be buggy.
I have noticed that on some hosts i start to get "First network error" in log right after proxy restart, so i decided to sniff some trafiic.
From dumps i see that right after restart poller starts to increase number of requests per packet as intended. And then suddenly ICMP ping check to same host with 5 packets kicks in. And right after that SNMP poller drops OIDs per packet to 1-3.
Probably it takes those ICMP checks in account and decides that only 5 OIDs were successful?
- duplicates
-
ZBX-8528 random lost UDP packets lead to not bulk snmp requests and as results increased CPU usage etc
- Closed