Hi everyone,
I’m currently facing an issue with KumoMTA where restarting the service takes an excessively long time to enumerate the spool, around 7-8 hours. This seems to be due to the high volume of emails in the queue, which is over 1-1.5 million messages.
Does anyone have any suggestions on how I can minimize the spool enumeration time during a service restart? Any tips, optimizations, or configuration tweaks that could help reduce this downtime would be greatly appreciated!
There are some improvements to that in the latest Dev branch releases, what version are you on?
Also what is the IOPS on your spool drive?
iops= 3000
how can i check the version?
/opt/kumomta/sbin/kumod --version
version = kumod 2024.07.31-19bc0670
3000 IOPS, so a single SSD?
yes
on a single SSD
You may want to do the math on your drive’s throughput vs the size of your spool, you may be up against the physical limits of the device. I’d also suggest trying the latest -dev and seeing if there’s some help there, but IIRC that’s mostly related to speeding up the shutdown vs the restart.
What does kcli top look like during the restart?
this is top output
spoolin pool is at 0%, so there’s no messages loading.
I take it the restart is long over?
That is just now?
As in those lines are still populating?
yes
yes they are
How does your disk usage look in iostat?
