Currently in the worst case retry interval is:
[0.1, 0.2, 0.4, 0.8, 1.6, 3.2, 6.4, 10, 10, 10]
This may cause some neutron DB action fail, and result an
API 500/40X error, for instance, the IP allocation get a
large failure chance with such config.
After this change, the worst case retry interval will look
something like this:
[0.5, 1, 2, 4, 8, 10, 10, ... 10]
So, for the new value 20 and 0.5, in the worst case the total
time cost will be 165.5s. For the old config, it's 42.7s. It's
not too radical for the new retry input params. And we get a
higher rate of API success.
Reviewed: https:/ /review. openstack. org/583527 /git.openstack. org/cgit/ openstack/ neutron- lib/commit/ ?id=ed62a06a536 a39b1236bd4dcad b2dd13cc2c4b4b
Committed: https:/
Submitter: Zuul
Branch: master
commit ed62a06a536a39b 1236bd4dcadb2dd 13cc2c4b4b
Author: LIU Yulong <email address hidden>
Date: Mon Jul 16 09:56:47 2018 +0800
Make the DB retry mechanism more robust
Currently in the worst case retry interval is:
[0.1, 0.2, 0.4, 0.8, 1.6, 3.2, 6.4, 10, 10, 10]
This may cause some neutron DB action fail, and result an
API 500/40X error, for instance, the IP allocation get a
large failure chance with such config.
After this change, the worst case retry interval will look
something like this:
[0.5, 1, 2, 4, 8, 10, 10, ... 10]
So, for the new value 20 and 0.5, in the worst case the total
time cost will be 165.5s. For the old config, it's 42.7s. It's
not too radical for the new retry input params. And we get a
higher rate of API success.
Change-Id: I5ad139bdfb3ae1 25658b36d05f85f 139a1b47bee
Closes-Bug: #1777968