#1214 Server status reset in failover might cause an infinite loop
Closed: Fixed None Opened 7 years ago by jhrozek.

A server status in fail over is reset from "not working" to "unknown" after an interval, which is currently hardcoded to 30 seconds. This might be a problem when it takes a long time for individual server tries.

For example, the default krb5_child timeout is 15 seconds, so when there is more than 2 servers in the fail over chain, the first server is always reset to the "default" state when the back end completes a full cycle over the servers.

We should modify the fail over (or the be_fo_ wrapper) to only allow one full cycle over the server list.

Fields changed

milestone: NEEDS_TRIAGE => SSSD 1.8.1 (LTM)
owner: somebody => jhrozek
priority: major => blocker

Fields changed

status: new => assigned

Fixed by:
- master
- b300bad
- 606d2d0
- sssd-1-8
- b3cd4ec
- 8e59877

component: SSSD => Failover
patch: 0 => 1
resolution: => fixed
status: assigned => closed

Metadata Update from @jhrozek:
- Issue assigned to jhrozek
- Issue set to the milestone: SSSD 1.8.1 (LTM)

2 years ago

Login to comment on this ticket.