VM’s sometimes failing to get their cloud-init data
Bug #1427642 reported by
Bogdan Dobrelya
This bug report is a duplicate of:
Bug #1411732: GRE - fragmentation packages on big size MTU.
Edit
Remove
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Fuel for OpenStack |
Confirmed
|
Medium
|
Fuel Library (Deprecated) | ||
| 6.0.x |
Confirmed
|
Medium
|
Fuel Library (Deprecated) | ||
Bug Description
VM’s were failing to get their cloud-init data about 30% of the time. Digging into this more, it appeared to be caused by packet drops/timeouts.
The w/a is to adjust the dhcp-agent to set the VM’s with a MTU of 1454 (Option 26).
Ideally, Fuel should enable Jumbo frames for storage and we should look at optimal settings for the management networks and underlay network including OVS.
| Changed in fuel: | |
| importance: | Undecided → Medium |
| milestone: | none → 6.1 |
| assignee: | nobody → Fuel Library Team (fuel-library) |
Revision history for this message
| Vladimir Grujic (hyperbaba) wrote | #1 |
Revision history for this message
| Vladimir Kuklin (vkuklin) wrote | #2 |
| Changed in fuel: | |
| status: | New → Incomplete |
| Changed in fuel: | |
| status: | Incomplete → Confirmed |
| tags: | added: customer-found |
Revision history for this message
| Andrew Woodward (xarses) wrote | #3 |
| tags: | added: low-hanging-fruit |
To post a comment you must log in.
Jumbo frames would be nice on all networks. This will reslove the packet fragmentation issues on running instances caused by gre tunnels over 1500 mtu links with their own 1500 mtu.