We need to periodically test the underlying ceph subnets for lost packets

Bug #1966322 reported by Steven Parker
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Ceph OSD Charm
New
Undecided
Unassigned

Bug Description

Due to an operation issue with a customer with a bad switch port we found it extremely hard to pinpoint the issue as related to networking issues. Existing ceph diagnostics and logs did not allow us to quickly pin point this issue.

I would request a new subordinate to do periodic fast fire ping tests using large MTU packets such as this:
   ping x.x.x.x -s 8900 -i 0.1

   28 packets transmitted, 12 received, 57% packet loss, time 2804ms

Pings could be done to three or so IPs that could be shared by the subordinate relations.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.