since few days, my zabbix-server decided to… take a nap.
After bashing some Windows Mobile devices, i took a coffee trying to understand what the funk was going on.
I were able to login to the interface, but the process was not started, as stated by the dashboard.
systemctl start zabbix-server did not ignite the whole gears, rebooting the system did not change a thing. So… runt through the log.
systemd is not capable to find the pid to babysit, therefore a tail /var/log/zabbix/zabbix_server.log led me to the issue…
My solution is “stupid” and working, but i would like to summon @mrmarkuz which is the maintainer of the package and @Andy_Wismer as the resident Zabbix-guru of the forum.
cp /etc/zabbix/zabbix_server.conf /etc/zabbix/zabbix_server_bck.conf
for create a copy of the config file, than i changed the section of /etc/zabbix/zabbix_server.conf
### Option: CacheSize
# Size of configuration cache, in bytes.
# Shared memory size for storing host, item and trigger data.
#
# Mandatory: no
# Range: 128K-8G
# Default:
# CacheSize=8M
adding
CacheSize=128M
at the bottom
Questions are:
@mrmarkuz the change to the config file I suppose won’t survive to any signal-event related to zabbix_server, correct?
@Andy_Wismer for what i can understand, there was no defined cache size for the system. 128 is quite arbitrary and 16 times the default size. The installation is a VM with 4gb of ram hosting guacamole (still not used), zabbix and backup of the system… swap seem not used. I’m open for suggestions about a “reasonable” cache size, hint that could be helpful for other zabbix wannabe-admins better than me.
I so have never needed to define this on any of the over 30 Zabbix i maintain… (That’s why I’m wondering).
I will check and report, but need to go out quickly…
At home my Zabbix database is huge 12-16 GB… At my clients it’s usually 4-10 GB, depending on size of LAN, devices. Mine is large due to testing a lot!
This is related to PostGRE SQL and not to Zabbix. Anyway…
Currently the system have this version of PostGRE SQL
postgresql-9.2.24-7.el7_9.x86_64
PGTune do not refer to 9,2 but only to 9.5. I inserted the data related to my installation, which is a VM and not bare metal. This the output provided.
# DB Version: 9.5
# OS Type: linux
# DB Type: web
# Total Memory (RAM): 4 GB
# CPUs num: 2
# Data Storage: hdd
max_connections = 200
shared_buffers = 1GB
effective_cache_size = 3GB
maintenance_work_mem = 256MB
checkpoint_completion_target = 0.9
wal_buffers = 16MB
default_statistics_target = 100
random_page_cost = 4
effective_io_concurrency = 2
work_mem = 5242kB
min_wal_size = 1GB
max_wal_size = 4GB
max_worker_processes = 2
Going full cache (4GB) seem an overkill to me, i was asking myself if 128M should not be a bit too much aggressive…
