Active agent items delayed - false alerts (Host unreachable)

[dimir]

Can you ensure the time on affected agent is in sync?

[nerijus]

I am getting the same problem with a few active Windows agents. I have lots of such messages in Dashboard with negative duration:
Zabbix agent on Example is unreachable for 5 minutes Duration: - 1m 32s
The time on server and client is the same.
It started to happen after server upgrade to 4.0.3.

[LenR]

We upgraded to 4.0.1 Sunday and are having similar problems. Only agent(active) items and only get false alerts on Windows systems. The proxy is about half Linux, but all alerts have been windows. I have a fuzzytime test, changing the system.localtime to Zabbix active fixes the problem, but I can't get to the agent port on some systems, so of course, those fail.

I have support and have opened a case, but we had to solve a "Duplicate entry" log error which doesn't seem to have changed the queue problem.

[LenR]

My debug log also shows big time offset with agent active items, always windows agent, but apparently not all of them.

However, running this on the proxy shows the time is in sync

zabbix_get -s some-system -k system.localtime && date +%s
1546950358
1546950358

Like others, we didn't have this problem prior to 4.0, then had it on 10's of systems immediately after the upgrade.

[LenR]

Another update, these seem to be coming from systems with high uptime, like +200 days. I just had the zabbix agent restarted on one of them and it appears to have fixed the problem.

FromL
log_client_timediff(): timestamp from json 1546912645 seconds and 450148177 nanosecond, delta time from json 1404 seconds and 627193920 nanosecond

To:
log_client_timediff(): timestamp from json 1546960731 seconds and 955323335 nanosecond, delta time from json 0 seconds and
192417745 nanosecond

[LenR]

It's this: https://support.zabbix.com/browse/ZBX-15301

[mrogers-9898]

Yep, that was positively it for me, updating to 4.0.4rc1 completely solved my agent issues.