Description of problem:

Existing dmeventd logic uses a single lvm2 thread to process all repair commands. Idea here is the usually disasters do not happen all at the same time. However 'standalone' mirror repair may takes hours to replicate mirror image and during this time dmeventd is not able to do any other action.

Now we have lvmpolld to offload most of its work to external process and making dmeventd responsible only for quick urgent repair again and leaving long running task again being executed bug 'normal' daemons (without memlocking issues like dmeventd is using).

This should also lead to better command repair split - where only 'urgent-repair' parts are running within command context - and later i.e. mirror leg addition and resynchronization are fully under lvmpolld control.

It's important dmeventd quickly returns back to monitoring and is able to handle other incoming events.

Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info: