I tested setting the expiration to 1 hour and after restarting keystone and memcache everything works as expected.
About 30 minutes later when I try to spawn 30 concurrent instances keystone-all runs at 100% cpu and 90% of instances fail to spawn.
I switch back to using driver = keystone.token.backends.sql.Token and can spawn 30 instances concurrently multiple times without issues.
keystone-all never uses more than 20% cpu.
We have 2 keystone servers behind LB and 2 memcache servers running local to each keystone. Each keystone has a entry point to each of the memcache servers:
Do you expect the workaround to function as if the patch was applied or do you expect the patch to resolve the issue and the workaround as not something that is a valid test?
I tested setting the expiration to 1 hour and after restarting keystone and memcache everything works as expected.
About 30 minutes later when I try to spawn 30 concurrent instances keystone-all runs at 100% cpu and 90% of instances fail to spawn.
I switch back to using driver = keystone. token.backends. sql.Token and can spawn 30 instances concurrently multiple times without issues.
keystone-all never uses more than 20% cpu.
We have 2 keystone servers behind LB and 2 memcache servers running local to each keystone. Each keystone has a entry point to each of the memcache servers:
[memcache] 109:11211, 10.34.124. 142:11211
servers = 10.34.124.
Do you expect the workaround to function as if the patch was applied or do you expect the patch to resolve the issue and the workaround as not something that is a valid test?
Thanks
JC