Network/UpdateDeployments taking over five minutes to run

Bug #1626673 reported by Ben Nemec
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
tripleo
Fix Released
High
Unassigned

Bug Description

See output from heat-deploy-times:

2016-09-22 13:28:01.000 | UpdateDeployment 356.0
2016-09-22 13:28:01.000 | NetworkDeployment 351.0

Nearly six minutes for the network and update deployments. From what I can see, this is basically the time it takes for a node to reboot and apply os-net-config after it finishes deploying. In a virt environment like CI that should not take 6 minutes. More like 30 seconds.

From watching the console log during a deployment, the problem appears to be that during boot we are trying to DHCP all of the nics on the system, including all the isolated networks that don't have DHCP. This takes 5 minutes to time out, hence the 5 minute+ deployment times.

I suspect it's actually costing us 10 minutes per deployment because I think the deploy ramdisk and the actual system boot are both waiting this 5 minutes. While functionally this isn't a big deal, wasting 10 minutes per deployment this way is ridiculous and we should find some way to stop it.

Revision history for this message
Ben Nemec (bnemec) wrote :
Changed in tripleo:
status: Triaged → Fix Committed
milestone: none → newton-rc2
Changed in tripleo:
status: Fix Committed → Fix Released
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.