I am suppose to configure pmm2 QAN dashboard for Production. I have enabled both perfomance_schema and slow_log tables. But I am confusing in between these two options for --query-source option.
Which one I should use and which one would be better to identify the exact slow queries?
If I use --query-source=perfschema, what happens?
If I use --query-source=slowlog, what happens?
Can I get proper slow queries if I use either perfschema or slowlog option?
Can somebody help me any disadvantages or advantages?
If you use slow log, then PMM will be doing a tail -f on your slow log, reading query data as it is written to the slow log and processing it.
If you use perf-schema, then PMM will query certain performance_schema tables every few seconds and grab the data and process it.
The main reason for these two options is that if you are using, say AWS RDS, or some other DBaaS, then you don’t have access to the disk where slow log files exist. Thus, the only option is perfschema. If this is not a DBaaS situation, I would stick with slow-log.
NO NO NO NO NO!!! NO NO NO!! Do NOT enable slow log as a table! NO NO NO! Keep this as a file on disk. You will suffer horrible performance issues by having the slow log as a table.
Thanks for the help.
Do I need to enable both perfschema with disable table stats and slowlog with enable table stats, will it be fine. As I already enabled the slow_log table in the prod servers.
Can I add like below?
pmm-admin add mysql --query-source=perfschema --username=pmm --password=pmm --disable-tablestats – metrics monitoring
Thanks for the help.
Last one, do we have any option like read from slow_log table instead of slow log file.? As I am already enabled the slow/_log table in the production.