ceph update procedure

[RobFantini]

In another ceph thread:

I'd in general recommend a more cautious approach, restarting all those daemons at once can lead to issues, especially if there are unexpected problems with the new version.

we want to be as cautions as possible on ceph cluster upgrades.


this is how we currently do ceph upgrades , and I am not sure that it is cautious enough. please advise


1- do mon systems 1st

2- restart services by

systemctl try-reload-or-restart pvestatd.service pvedaemon.service

systemctl restart osd.target

systemctl restart ceph-mon.target

3- make sure ceph healthy then do next node. use
watch ceph -s
OR: check ceph status at pve web page. let health get normal before doing next node

 

[RobFantini]

the update procedure for ceph changed this time [ to 15.2.13 ]

this no longer works

# systemctl restart osd.target
Failed to restart osd.target: Unit osd.target not found.

so used the following. these are quick notes not well formatted

# 1
apt-get update && apt-get full-upgrade


# 2-  restart services

systemctl try-reload-or-restart pvestatd.service pvedaemon.service
systemctl restart ceph-mon.target

# type this line then press tab
systemctl restart ceph-mgr@
#    will do this

systemctl restart ceph-mgr@pve10.service

# this is new, need to restart one osd at a time.  that is safer then all at once.  use 'df' or check pve wepage to know the osd Numbers to restart

systemctl restart ceph-osd@4.service
systemctl restart ceph-osd@10.service
systemctl restart ceph-osd@25.service
systemctl restart ceph-osd@0.service


# 3- make sure ceph healthy then do next node.
watch ceph -s   ###  When all set do next node
                            ###  OR: check ceph status at pve web page.   let health get normal  b4 doing next node

 

[spirit]

Code:
# systemctl restart osd.target
Failed to restart osd.target: Unit osd.target not found.

systemctl restart ceph-osd.target

 

[RobFantini]

systemctl restart ceph-osd.target

thanks, wil use that next time.