[ZBX-13631] at startup Zabbix-3.4.4 uses zabbix_server_alerter.sock in /tmp before reading SocketDir from config Created: 2018 Mar 20  Updated: 2018 Mar 22  Resolved: 2018 Mar 22

Status: Closed
Project: ZABBIX BUGS AND ISSUES
Component/s: Server (S)
Affects Version/s: 3.4.6, 3.4.7
Fix Version/s: None

Type: Incident report Priority: Minor
Reporter: Altjo Raap Assignee: Unassigned
Resolution: Won't fix Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified


 Description   

Firstly, i am new here, so if you have any tips on entering bugs!

It may happen in all 3.x versions, i do not know.

I have two Zabbix environments on one system, which i upgraded, one from
3.2 to 3.4.4 and one from 2.4.2 to 3.4.4.
I was unaware of SocketDir for the alerting socket, and the default to create them in /tmp.
I had trouble starting the second environment and found out this socket thing.
I changed the second environment to point SocketDir=<some_where_else>.
Starting zabbix_server in the second environment gave alerts in the first.
So, it looks like the alerting routines are connecting to the default socket in /tmp before the config with SocketDir=<some_where_else> is set.

Steps to reproduce:
Create two seperate Zabbix environments on one system (1 and 2 for reference)
Startup env 1 (This will use /tmp for alerting socket)
In env. 2 put seperate folder for SocketDir in config.
In env. 2 take care of triggering alerts on a regular basis.

Stop env. 2.
Start env. 2

  1. Changes in configuration...
  2. Navigate to screen title...
  3. Click on screen element...
  4. ...

Result:
I get alerts from env. 2 in env. 1

Expected:
No Alerts from env. 2 in Env. 1

 Comments   
Comment by Andris Zeila [ 2018 Mar 20 ]

That's sounds really impossible for two reasons:

  1. zbx_ipc_service_init_env() is called after conf file has been loaded with zbx_load_config() function and before any of zabbix processes are forked
  2. ipc services will refuse connections initiated by processes started by another zabbix server.

Are you sure it's env2 generating those events, not env1 ?

Comment by Andris Zeila [ 2018 Mar 20 ]

Actually if we are talking about notifications configured in actions then the alert sent by alert manager to alerter contains all data, so even if somehow it was sent to another zabbix instance, the alert itself would be delivered by using the first instance settings.

Are you sure it's not something else (using one database or like)?

Comment by Altjo Raap [ 2018 Mar 22 ]

It appears to be we were missinterpreting the message.
We seem to have something going wrong in the discovery of disks on a windows machine.
The message is about diskspace of an F: and an G: drive which are not there (anymore?), the message never went away.
We have been playing with "Update interval" now, and the alert/problem is gone.
Lets hope it stays gone now.
So, this just came to light at the time of the upgrade and restarting the 2 zabbix environments.
If this does not resolve in the long term i have to create a ticket on vfs.fs.discovery key in zabbix agent.
This ticket can be closed now

Comment by Glebs Ivanovskis (Inactive) [ 2018 Mar 22 ]

Thank you for the update! Closing.

Generated at Fri Apr 04 09:27:49 EEST 2025 using Jira 9.12.4#9120004-sha1:625303b708afdb767e17cb2838290c41888e9ff0.