Hi Jan, On 2/10/23 7:23 AM, Jan Marek wrote:
osd basic osd_memory_target 1073741824
Any idea where this is coming from? As Radek mentioned, increasing the priority_cache debug level to 10+ should help diagnose the issue. That assert means we are assigning more memory in balance_priority than is available to be assigned. I suspect that as we loop over caches in balance_priority, we must end up allowing the aggregate fair_share value of all caches to exceed the remaining mem_avail. There previously was a bug where we asserted too early if even just assigning a single "chunk" to each cache pushed us over the limit, but that was fixed here: https://github.com/ceph/ceph/pull/27763 Mark -- Best Regards, Mark Nelson Head of R&D (USA) Clyso GmbH p: +49 89 21552391 12 a: Loristraße 8 | 80335 München | Germany w: https://clyso.com | e: mark.nelson@clyso.com We are hiring: https://www.clyso.com/jobs/