Victoriametrics cpu usage looks much higher than prometheus

Percona Monitoring and Management (PMM) PMM 2.x
9

Hi @ nurlan thanks for your replay

I’ve tried checking the slowlog, but it’s really a bit difficult… I didn’t find any unusual statements, or at least I found that using pt-query-digest to analyse these slow log files was fine. I am using Percona 5.7.29/32

Could the agent be more robust, and if it encounters a log it can’t parse, skip it and output the warning to the log. I feel better than rebooting

Here are my slow query related parameters

log_timestamps                              = SYSTEM
log_output                                  = file
slow-query-log-file                         = /data/mysql_3306/logs/slow.log
slow_query_log                              = ON
long_query_time                             = 0
log_slow_rate_limit                         = 100
log_slow_rate_type                          = query
log_slow_verbosity                          = full
log_slow_admin_statements                   = ON
log_slow_slave_statements                   = ON
slow_query_log_always_write_time            = 1
slow_query_log_use_global_control           = all

I think even if there is a truncated query statement, the agent should be able to handle such exceptions instead of restarting

here’s the slow log file and /var/log/message. Hope this helps for debug
https://www.swisstransfer.com/d/07d6ecbb-1715-42b2-912e-7a8fac6a044b

1 Like