With CentOS/Rocky 7-8 I’ve observed unexpected usage of swap when there is plenty of physmem available.
Swap IMHO is a relic of a time when RAM capacities were lower and much more expensive. In years beginning with a 2, and with Ceph explicitly, I assert that swap should never be enabled during day to day operation. The RAM you need depends in part on the media you’re using, and how many per node, but most likely you can and should disable your swap. > > Hello, > We have a 6-node ceph cluster, all of them have osd running and 3 of them > (ceph-1 to ceph-3 )also has the ceph-mgr and ceph-mon. Here is the detailed > configuration of each node (swap on ceph-1 to ceph-3 has been disabled after > the alarm): > > # ceph-1 free -h > total used free shared buff/cache > available > Mem: 187Gi 38Gi 5.4Gi 4.1Gi 143Gi > 142Gi > Swap: 0B 0B 0B > # ceph-2 free -h > total used free shared buff/cache > available > Mem: 187Gi 49Gi 2.6Gi 4.0Gi 135Gi > 132Gi > Swap: 0B 0B 0B > # ceph-3 free -h > total used free shared buff/cache > available > Mem: 187Gi 37Gi 4.6Gi 4.0Gi 145Gi > 144Gi > Swap: 0B 0B 0B > # ceph-4 free -h > total used free shared buff/cache > available > Mem: 251Gi 31Gi 8.3Gi 231Mi 211Gi > 217Gi > Swap: 124Gi 3.8Gi 121Gi > # ceph-5 free -h > total used free shared buff/cache > available > Mem: 251Gi 32Gi 14Gi 135Mi 204Gi > 216Gi > Swap: 124Gi 4.0Gi 121Gi > # ceph-6 free -h > total used free shared buff/cache > available > Mem: 251Gi 30Gi 16Gi 145Mi 204Gi > 218Gi > Swap: 124Gi 4.0Gi 121Gi > > We have configured swap space on all of them, for ceph-mgr nodes, we have 8G > swap space and 128G swap configured for osd nodes, and our zabbix has > monitored a swap over 50% usage for ceph-1 to ceph-3, but our available space > are still around 140G against the total 187G. Just wondering whether the swap > space is necessary when we have lots of memory available? > > Thanks very much for your answering. > _______________________________________________ > ceph-users mailing list -- ceph-users@ceph.io > To unsubscribe send an email to ceph-users-le...@ceph.io _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
