Hi everyone, my check_mk on virtual machine (hypervisor ProxMox ve 5.1) has all services and hosts in stale state. It happens since I start to add new hosts. I have above 1800 hosts. I gave to system 32 of 54 cores, processors - 56 x Intel® Xeon® CPU E5-2680 v4 @ 2.40GHz (2 Sockets) x2
By taking a look in htop, ive not seen anything suspicious.
my problem looks like on screenshot 2
Could anyone help me with this? Or give me a right direction to find a solution?
Hi Alexander
is this a new installation? If not, did something change on the config side?
Which version of 1.4.0 is it? Core (i guess it is) or Enterprise?
In general stale and/or hosts mean that the core is not able to compute the host/service check-queries in a sufficient time (i.e. “overcharged”).
What is host/service check interval?
Which checks do you have, i.e. SNMP, agent, SSH…
Also imprtant to know:
I do not know Proxmox, but in general in a VM all CPU cycles are virtualized, so for example in vSphere you can preserve a certain amout of CPU power directly for a VM to compensate this.
BR
Am Di., 29. Jan. 2019 um 14:11 Uhr schrieb Александр Чекалов chekalov@samrct.ru:
Hi everyone, my check_mk on virtual machine (hypervisor ProxMox ve 5.1) has all services and hosts in stale state. It happens since I start to add new hosts. I have above 1800 hosts. I gave to system 32 of 54 cores, processors - 56 x Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz (2 Sockets) x2
By taking a look in htop, ive not seen anything suspicious.
my problem looks like on screenshot 2
Could anyone help me with this? Or give me a right direction to find a solution?
checkmk-en mailing list
checkmk-en@lists.mathias-kettner.de
Manage your subscription or unsubscribe
https://lists.mathias-kettner.de/cgi-bin/mailman/listinfo/checkmk-en
Hi Alexander
thanks for your quick reply.
If it says Raw on the top left, then its the core or CRE which has the free Nagios core to compute all queries.
Snmp by design is the slowest method to get a host information so you have to look into your monitoring config.
Less tahtn 5 minutes per host or service check are useless. then go to your ruleset and search for “snmp”.
Inspect all values, especially timeout settings.The snippet you provided is just for Multisite (the interface). You need to tune the monitoring.
What you could try first hand is to switch off SNMP for all hosts (i.e. ping only)and see if staleness disappears.
BR
Am Di., 29. Jan. 2019 um 15:01 Uhr schrieb Александр Чекалов chekalov@samrct.ru:
Its not a new installation, yes, ive start add new hosts to database. Was approx. 56 hosts, I add approx. 1700 more.
Its RAW 1.4.0p35, I guess core.
If there is not valid timers which you asking for, could you tell me where I can find it?
I have only snmp and ping.
About proxmox, I tried to share all of my calculate power with that vm, but it doesn’t work.
I try to disable notifications and its works. Maybe I have insufficient number of monitoring worker processes? Any idea?
From: Thomas Wittmann [mailto:tom.teel@gmail.com]
Sent: Tuesday, January 29, 2019 5:48 PM
To: Александр Чекалов <chekalov@samrct.ru>
Cc: checkmk-en@lists.mathias-kettner.de checkmk-en@lists.mathias-kettner.de
Subject: Re: [Check_mk (english)] ProxMox VM cmk 1.4.0 stale hosts
Hi Alexander
is this a new installation? If not, did something change on the config side?
Which version of 1.4.0 is it? Core (i guess it is) or Enterprise?
In general stale and/or hosts mean that the core is not able to compute the host/service check-queries in a sufficient time (i.e. “overcharged”).
What is host/service check interval?
Which checks do you have, i.e. SNMP, agent, SSH…
Also imprtant to know:
I do not know Proxmox, but in general in a VM all CPU cycles are virtualized, so for example in vSphere you can preserve a certain amout of CPU power directly for a VM to compensate this.
BR
Am Di., 29. Jan. 2019 um 14:11 Uhr schrieb Александр Чекалов chekalov@samrct.ru:
Hi everyone, my check_mk on virtual machine (hypervisor ProxMox ve 5.1) has all services and hosts in stale state. It happens since I start to add new hosts. I have above 1800 hosts. I gave to system 32 of 54 cores, processors - 56 x Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz (2 Sockets) x2
By taking a look in htop, ive not seen anything suspicious.
my problem looks like on screenshot 2
Could anyone help me with this? Or give me a right direction to find a solution?
checkmk-en mailing list
checkmk-en@lists.mathias-kettner.de
Manage your subscription or unsubscribe
https://lists.mathias-kettner.de/cgi-bin/mailman/listinfo/checkmk-en
Hi Alexander
please read carefully. I suggested to disable SNMP just to test if staleness disappears, which could be an indicator that the Nagios core is overloaded.
You still miss the information about your config, Linux etc
This is like looking into a glass bowl
BR
Am Mi., 30. Jan. 2019 um 06:35 Uhr schrieb Александр Чекалов chekalov@samrct.ru:
Thanks for your advices, Thomas. But snmp is very useful for my work. I need to know if interfaces down or up and most important I need to check the MAC addresses of hosts, to actualize ip plan and schema. So disable snmp is not an option for me. How do you think, why disabling notifications is worked for me?
From: Thomas Wittmann [mailto:tom.teel@gmail.com]
Sent: Tuesday, January 29, 2019 6:10 PM
To: Александр Чекалов chekalov@samrct.ru; checkmk-en@lists.mathias-kettner.de checkmk-en@lists.mathias-kettner.de
Subject: Re: [Check_mk (english)] ProxMox VM cmk 1.4.0 stale hosts
Hi Alexander
thanks for your quick reply.
If it says Raw on the top left, then its the core or CRE which has the free Nagios core to compute all queries.
Snmp by design is the slowest method to get a host information so you have to look into your monitoring config.
Less tahtn 5 minutes per host or service check are useless. then go to your ruleset and search for “snmp”.
Inspect all values, especially timeout settings.The snippet you provided is just for Multisite (the interface). You need to tune the monitoring.
What you could try first hand is to switch off SNMP for all hosts (i.e. ping only)and see if staleness disappears.
BR
Am Di., 29. Jan. 2019 um 15:01 Uhr schrieb Александр Чекалов chekalov@samrct.ru:
Its not a new installation, yes, ive start add new hosts to database. Was approx. 56 hosts, I add approx. 1700 more.
Its RAW 1.4.0p35, I guess core.
If there is not valid timers which you asking for, could you tell me where I can find it?
I have only snmp and ping.
About proxmox, I tried to share all of my calculate power with that vm, but it doesn’t work.
I try to disable notifications and its works. Maybe I have insufficient number of monitoring worker processes? Any idea?
From: Thomas Wittmann [mailto:tom.teel@gmail.com]
Sent: Tuesday, January 29, 2019 5:48 PM
To: Александр Чекалов <chekalov@samrct.ru>
Cc: checkmk-en@lists.mathias-kettner.de checkmk-en@lists.mathias-kettner.de
Subject: Re: [Check_mk (english)] ProxMox VM cmk 1.4.0 stale hosts
Hi Alexander
is this a new installation? If not, did something change on the config side?
Which version of 1.4.0 is it? Core (i guess it is) or Enterprise?
In general stale and/or hosts mean that the core is not able to compute the host/service check-queries in a sufficient time (i.e. “overcharged”).
What is host/service check interval?
Which checks do you have, i.e. SNMP, agent, SSH…
Also imprtant to know:
I do not know Proxmox, but in general in a VM all CPU cycles are virtualized, so for example in vSphere you can preserve a certain amout of CPU power directly for a VM to compensate this.
BR
Am Di., 29. Jan. 2019 um 14:11 Uhr schrieb Александр Чекалов chekalov@samrct.ru:
Hi everyone, my check_mk on virtual machine (hypervisor ProxMox ve 5.1) has all services and hosts in stale state. It happens since I start to add new hosts. I have above 1800 hosts. I gave to system 32 of 54 cores, processors - 56 x Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz (2 Sockets) x2
By taking a look in htop, ive not seen anything suspicious.
my problem looks like on screenshot 2
Could anyone help me with this? Or give me a right direction to find a solution?
checkmk-en mailing list
checkmk-en@lists.mathias-kettner.de
Manage your subscription or unsubscribe
https://lists.mathias-kettner.de/cgi-bin/mailman/listinfo/checkmk-en