zabbix for 25k devices

[giotto69]

Hi guys
I've been asked to estimate the costs to deploy a zabbix solution to poll 25000 catalyst devices in terms of hardware usage and solution setup.
The request is to get via SNMP bandwidth usage IN and OUT every 5 minutes.
The number of ports involved is around 300K

According to your experience, can zabbix manage this load?

thanks you!
Giorgio

[richlv]

according to a quick calculation (hopefully correct one), that should be 2000 new values per second. if your database can handle some load, that shouldn't be an issue

[giotto69]

Thanks richlv,
for my quotation I would need an hardware estimation too.
Is there a target load for one single zabbix poller, or some kind of reference to make such an estimation?
thanks

[richlv]

a poller count alone doesn't tell you anything, it would be mostly be based on nvps anyway. i don't have any quick hw suggestions right now, just make sure to have lots of ram and fast disks for the database

[Colttt]

use SSD and 12GB RAM..

and i think use one server for the datebase and one server for the zabbix-server..

[alp]

Hi guys
I've been asked to estimate the costs to deploy a zabbix solution to poll 25000 catalyst devices in terms of hardware usage and solution setup.

How you can start server?
I have 16000 switches with snmp items from standart template and my zabbix crashed after start =\

http://pastie.org/3919444 log of crash

mon3 / # sysctl -a | grep shm
kernel.grsecurity.chroot_deny_shmat = 1
kernel.shmmax = 204800000
kernel.shmall = 40960000
kernel.shmmni = 4096

In zabbix_server.conf i have:

CacheSize=60M
HistoryCacheSize=16M
HistoryTextCacheSize=2M
TrendCacheSize=64M

StartPollers=50
StartPollersUnreachable=10
StartTrappers=50
StartPingers=10

When i set CacheSize=1G server crashed anyway:
30780:20120514:130456.483 cannot allocate shared memory of size 912680551: [22] Invalid argument
30780:20120514:130456.483 cannot allocate shared memory for configuration cache

[Colttt]

use a smaller cachesize!!

[alp]

use a smaller cachesize!!

Thanks, captain obvious!

I'm write about two situation and all of it give me error!

[Colttt]

hmm maybe a bug??
do you report it?

[richlv]

so you tried with 60m and 1g, but never anything inbetween ?

[alp]

so you tried with 60m and 1g, but never anything inbetween ?

I trying with any limits in many tryes.

[frankymryao]

we use:

Server: Red Hat 4.1.2-46,16 cores, 32G.
3 Proxy:Red Hat 4.1.2-46, 4 cores, 8G,
Oracle server: Red Hat 4.1.2-46, 16 cores, 16G, 10 SATA disk, Oracle 11.2.0.3.0

Handle 1600vps and has 30%~40% free capacity.

But, I have patched some bad codes in zabbix c codes. Before patching, zabbix is not as stable as it would to be and now it works well.

BTW, we use zabbix 1.8.8