Feature Request – Locked Query Tracking And Persistent Alerts

[suhrick]

1) So far in Monyog the only way I have found to get a history of who-locked-who is under “Real Time”.  However, I need to start Real Time fresh each morning, the session doesn’t continue to gather data when the browser page for Real Time isn’t up.  This allows me to get an idea of who is locking who over the course of a single work-day, but I need to close down my laptop each night, needed for off-hour support.  So I have no idea of what does on during the night.  Can the locking tracking be added to Query Analyzer or another persistent tool?

 

2) Monyog monitors send emails only three times; when warning threshold is crossed, when alert threshold is crossed and when things become stable.  For replication this isn’t really useful, emails can get overlooked in the mix.  It would be great if Monyog continued to end out emails regularly on the monitors the client desires.  So I would get “2,000 seconds behind”, followed by “3,000 seconds behind”, followed by “4,000 seconds behind”, continuing until I fix the damn thing…

 

Thanks,

 

Scott Uhrick

Teladoc DBA

[Anonymous]

Which MONyog version are you using?

 

Where is MONyog running? In your laptop or some other machine? If it’s your laptop then shutting it down for an hour will shut MONyog down as well, we can’t do anything. If it’s running on some other machine then you can just safely close the browser and Real-Time will still collect data for 3 days. We included this feature in MONyog 6.2. More details here http://blog.webyog.com/2014/10/10/monyog-mysql-monitor-6-2-released/

 

For your second question, 6.22 onward, MONyog has a feature “Notify Till Stable” which will notify you till a counter comes to a stable state based on a reminder interval time setting. You can read more about that feature here. http://blog.webyog.com/2014/12/04/monyog-mysql-monitor-6-22-released/

 

Does this help you?

[sqlmike]

Saurabh, What about a locked query alert?  I would like to know when there are X number of processes waiting on a lock resource and fire an alert.

[siddhant]

You can enable the alert for locked query in the Monitor Long Running Query option of Processlist Sniffer in MONyog. It would alert you whenever a query takes more than a specified time and the state of which is locked. To configure this please goto Edit server > Advanced settings > Sniffer > Enable sniffing? = Yes > Sniffer mode = Processlist > Monitor Long Running Query = Yes > Long running query time = (according to your requirement) > Monitor only locked queries = Check > Save all.

 

This would send you an alert in case the state of a query is locked and it exceeds the given time.

[sqlmike]

Thank you.  Are there plans to allow the Sniffer mode to be Performance Schema when enabling the Monitor Long Running Query?  A feature to alert on the number of blocked processes instead of just time?

[siddhant]

Hi,

 

The query will appear in Performance Schema database only when it has finished execution. So the user won’t be able to kill the query while it is running like for sniffer in processlist mode. We’ll consider your request for adding the notification for long running query for sniffer in Performance Schema mode.

[sqlmike]

Thank you for the reply.  I think in the long run the need to use the Performance Schema will win out over process list for collecting data, but I do see your point.   There really needs to be something separate for monitoring blockers by time and by count in my opinion. 

 

 

edit:

 

I’m a long time dba, but fairly new to mysql and something that I would really like to see in monyog is something like percona’s pmp-check-mysql-innodb that will allow alerts to fire for waiter_count or waiter_duration.  Detecting sleeping connections that are still holding locks and blocking is really important.  Note that I have no relationship with percona, just went looking for a blocker monitoring solution.

[Author]

Posts