I have very stable and performant Zabbix solution that has been upgraded over the last 5 years and working well until recently. This past weekend, MariaDB just crashed. I tried several things and finally restored the whole VM to a point prior to the crash and it crashed again within an hour with what looked like Innodb database corruption.
Thanks!
- To resolve I restored again and shut all Zabbix down and upgraded to Zabbix to 6.0.26 and MariaDB to 10.5.24 all on Ubuntu 20.04.
- I noticed tables were huge so tweaked down history, trends and auditing and ran housekeeper as well as did some database tunning.
- I then installed Percona tools so I could optimize tables while online, since the history_uint tables was 130GB and would take forever.
- pt-online-schema-change worked on some other tables like audit etc but would not run on history or history_uint due to missing primary keys.
- At any rate, I got everything running again and everything seems ok now, but I would really like to reclaim space on those huge history tables and in checkin Zabbix documentation it seems that maybe I missed something in my upgrade to 6 to migrate to Primary keys.
Thanks!
Comment