Hi all,
We have 3 proxies in a data center that has high latency to our Zabbix server (2.4). We are around 2,600 NVPS right now. We are using a proxy (3.2) which I've recompiled with ZBX_MAX_HRECORDS=2500 (default=1000).
Due to upstream outages, etc., its not uncommon for the one or more of the proxies to fall behind. Sometimes they recover, sometimes not. We've noticed the proxy_history table sometimes grows very large.
We currently run the default housekeeper setting on the proxy, however I'm thinking about disabling that and going with a partitioning strategy on proxy_history to better manage size, i.e. partitioning by day and only keeping maybe 3 days of history.
Question 1 -- What criteria does the housekeeper use to determine "old" data? Or am I missing where this can be defined? Is "old" an hour, a day, a week, a month?
Question 2 -- Is it safe to mess with the proxy_history table? The mysql partitioning howto doc seems to suggest that Zabbix is tolerant of this.
Thanks in advance.
We have 3 proxies in a data center that has high latency to our Zabbix server (2.4). We are around 2,600 NVPS right now. We are using a proxy (3.2) which I've recompiled with ZBX_MAX_HRECORDS=2500 (default=1000).
Due to upstream outages, etc., its not uncommon for the one or more of the proxies to fall behind. Sometimes they recover, sometimes not. We've noticed the proxy_history table sometimes grows very large.
We currently run the default housekeeper setting on the proxy, however I'm thinking about disabling that and going with a partitioning strategy on proxy_history to better manage size, i.e. partitioning by day and only keeping maybe 3 days of history.
Question 1 -- What criteria does the housekeeper use to determine "old" data? Or am I missing where this can be defined? Is "old" an hour, a day, a week, a month?
Question 2 -- Is it safe to mess with the proxy_history table? The mysql partitioning howto doc seems to suggest that Zabbix is tolerant of this.
Thanks in advance.
Comment