Looking at the manage output, this does not make sense to me. its not exactly down
. however the performance is pretty much unacceptable. my understanding here is that there was an arbitary value set for a timeout which causes the seeminly poor performance problem. i guess the primary issue is that master-master has no idea how many minions it will talk to and thus can't determine a proper timeout (given the syndic is simply relaying). however, you have the issue on the other side of the timeout is too short if its talking to a large number of the infrastructure. it would seem to make some sense to potentially have the master-master have an idea as to the number of syndics it talks to and make a better call from there. maybe returns back a number of minions attached and you can assign a timeout value to each minion.
the other thought maybe to allow a prefix of the syndic name as it might be useful to know where the minions are sitting (unders what syndic's control). maybe i should PR that :)
[root@saltmaster ~]# time salt-run manage.status
down:
- syndic
up:
- minion1
real 0m20.705s
user 0m0.603s
sys 0m0.139s
[root@saltmaster ~]# time salt \* test.ping
minion1:
True
real 0m20.594s
user 0m0.436s
sys 0m0.093s```