TLDR

Set the interval before warning to 14 days:

ceph config set global osd_deep_scrub_interval 1209600

Problem

My Ceph cluster runs on cheap hardware and large OSDs take a while to scrub so I’m seeing Ceph health warnings:

[WRN] PG_NOT_DEEP_SCRUBBED: 43 pgs not deep-scrubbed in time
    pg 11.71 not deep-scrubbed since 2022-11-20T23:06:43.607051+0100
    pg 7.78 not deep-scrubbed since 2022-11-21T18:39:17.628021+0100

Default configuration

The default interval before warning is seven days:

root@pve:~# ceph config show-with-defaults mgr.pve | grep osd_deep_scrub_interval
osd_deep_scrub_interval 604800.000000 default

604800 seconds = 7 days

Increase window to 14 days

I didn’t want to spend more hours per day scrubbing, so for me the solution was to increase the windows to 14 days (1209600 seconds).

ceph config set global osd_deep_scrub_interval 1209600

Other notes

You can also adjust the number of hours per day scrubbing happens and which days:

• osd_scrub_begin_hour
• osd_scrub_end_hour
• osd_scrub_begin_week_day
• osd_scrub_end_week_day