View Full Version : Active Checks Limited To 1 With Alpha10?
It seems that when running alpha10 (Linux 2.4 kernel from clients and server) I can poll from the server a large amount of standard Zabbix Agent entries, but the data does not seem to update for more than one active check per host when I choose Zabbix Agent (active). The one that does work using active does work much more reliably - as such it would be great to switch everything over to active as the server is starting to get killed (just barely keeping up with the queue) monitoring 400 servers and 30 items per server.
Am I doing something wrong (where the "Latest Data" only updates 1 active check when I change multiple entries to active) or is this an alpha limitation?
Thx,
Matt
No, I'm not aware of any problems with active checks in 1.1alpha10. Any warnings or errors in zabbix_agentd's log file?
Have you checked that zabbix_agentd is running?
My experience with running active checks (on FreeBSD 5.4) resulted in reporting 1 time and then crash when trying to update from the server.
No, I'm not aware of any problems with active checks in 1.1alpha10. Any warnings or errors in zabbix_agentd's log file?
I setup Zabbix completely using non-active checks until I started hitting performance problems where the queue was barely keeping up with the various items I had polling every 60 seconds or so.
It was at that point where I tried changing and creating new checks that were active. I assumed that all you did was change the type - if not please let me know. What I found is that the "latest data" page only shows the very 1st active check being updated - all the rest remain stale. When I increase the debug level on the client I see the client sending the active data but just for the very 1st active check- I never see it try and change any other checks. The server is Linux 2.4 (RedHat) - the clients are Solaris and Linux and both have the same problem where only the 1st active check updates and the rest of the active checks do not.
Pls let me know if you'd like me to send you anything in the database, the binaries, etc. for further debugging.
Thx,
Matt
Do you see ZABBIX agent receives full list of active checks from the server when you start the agent?
It appears after an agent restart it is only getting back the single active check that is non-stale, and not all active checks. Note some of the active checks I would like to use are UserParameters - I assume these work the same?.
031745:20050708:090507 zabbix_agentd 31745 started
031745:20050708:090507 In init_list()
031745:20050708:090507 In refresh_metrics()
031745:20050708:090507 get_active_checks: host[10.32.194.145] port[10051]
031745:20050708:090507 Sending [ZBX_GET_ACTIVE_CHECKS
linuxNY
]
031745:20050708:090507 Before read
031745:20050708:090508 Read [proccpu[MktShare]:60:0
]
031745:20050708:090508 In delete_all_metrics()
031745:20050708:090508 Parsed [proccpu[MktShare]:60:0]
031745:20050708:090508 Key [proccpu[MktShare]]
031745:20050708:090508 Refresh [60]
031745:20050708:090508 Lastlogsize [0]
031745:20050708:090508 In add check [proccpu[MktShare]]
031745:20050708:090508 linuxNY:proccpu[MktShare]
031745:20050708:090508 In send_value()
031745:20050708:090508 OK
031745:20050708:090508 Sleeping for 60 seconds
Please, give me a screenshot of items defined for the host. I'd like to see that there are really more than one active check defined. Thanks!
You can see the last check value for 2 of the 3 active checks is stale. (only one active check is updating, though the non-active std checks are all still working)
You may also find the database entry for the hosts useful:
# echo "SELECT * FROM items WHERE hostid='10408' AND type=7;" | /usr/bin/mysql zabbix
itemid type snmp_community snmp_oid snmp_port hostid description key_ delay history trends nextcheck lastvalue lastclock prevvalue status value_type trapper_hosts units multiplier delta prevorgvalue snmpv3_securityname snmpv3_securitylevel snmpv3_authpassphrase snmpv3_privpassphrase formula error lastlogsize
45915 7 public interfaces.ifTable.ifEntry.ifInOctets.1 161 10408 Free disk space on /var/maestro diskfree[/var/maestro] 60 90 365 1120856040 973029376.000000 1120855995 973029376.000000 0 0 B 1 0 NULL 01024 0
45917 7 public interfaces.ifTable.ifEntry.ifInOctets.1 161 10408 Free disk space on / diskfree[/] 60 90 365 0 5546369024.000000 1120832873 5546409984.000000 0 0 B 1 0 NULL 0 1024 0
33347 7 public interfaces.ifTable.ifEntry.ifInOctets.1 161 10408 CPU Utilization sar[busy] 10 90 365 1120832960 30.00 1120832950 59.44 0 0 0 0 NULL 0 Not supported by agent0
It was the NoTimeWait parameter - uncommenting it fixed the problem where agents would only receive the first active check.
NoTimeWait=1