Description of problem:
If connection between goferd and qdrouterd is lost (network issue, qdrouterd restart, whatever), goferd does reconnect attempts based on a retry scheme:

 retry in 10 seconds
 retry in 12 seconds
 retry in 14 seconds
 retry in 17 seconds
..
 retry in 106 seconds

If everything is running but goferd waits, say, 35 seconds for another connection attempt, and if one tries to install a package to this host, it would fail with 20s timeout ("is katello-agent running?"). That isn't too much robust..

Could be goferd retry scheme be with smaller delays between successive reconnect attempts? (i.e. starting from delay 2s, then 3s, then 4s etc. until 10s forever)? Or is there some particular reason for the current delays?


Version-Release number of selected component (if applicable):
gofer-2.6.2-2.el7sat.noarch
python-gofer-2.6.2-2.el7sat.noarch
python-gofer-qpid-2.6.2-2.el7sat.noarch


How reproducible:
100%


Steps to Reproduce:
All on Satellite:
1. service qdrouterd stop
2. sleep 73
3. service qdrouterd start
4. sleep 1
5. hammer content-host package install --content-host-id .. --organization-id 1 --packages sysstat

(ensure the sleep timing such that the host just logged "gofer.messaging.adapter.proton.connection:108 - retry in 20 seconds" before starting qdrouterd)


Actual results:
package install fails with 20s timeout


Expected results:
package install succeeds


Additional info: