[Rstplib-users] "Slow" reconfiguration in 802.1w when Bridge Priority is changed by management
Status: Alpha
Brought to you by:
ralex
From: <al...@nb...> - 2001-11-12 14:42:15
|
It looks like the reconfiguration of the RSTP domain works too slow when the Bridge priority is changed by management. Consider the simplest case: two Bridges are connected with one connection. Now on the Bridge, that is not currently RootBridge, let us decrease the Priority so, that it becomes RootBridge. I see, that the domain becomes stable only after 4 seconds; the previous designated port becomes a Backup, after that - Root. Here is a trace (sorry): 16:47:45 B8322 > get-st-pcfg BridgeId: 8000-008220000001 RootId: 8000-008120000001 p01 8001 Fwd 8000-008120000001 8000-008120000001 8001 R E p02 8002 Dis 8000-008120000001 8000-008220000001 8002 E p03 8003 Dis 8000-008120000001 8000-008220000001 8003 E p04 8004 Dis 8000-008120000001 8000-008220000001 8004 16:48:06 B8322 > set-br-prio 0x4000 Changed rstp bridge priority 16:48:27 B8322 > 16:48:27:brige B8322 became root - bpdu 16:48:27:rec(B8322-p01) => Backup 16:48:27:roletrns(B8322-p01): ROOT_PORT=>BLOCK_PORT 16:48:27:topoch (B8322-p01): TCACTIVE=>INIT 16:48:27:topoch (B8322-p01): INIT=>INACTIVE 16:48:27:sttrans (B8322-p01): FORWARDING=>DISCARDING 16:48:27:roletrns(B8322-p01): BLOCK_PORT=>BLOCKED_PORT 16:48:27:roletrns(B8322-p01): BLOCKED_PORT=>BACKUP_PORT 16:48:27:roletrns(B8322-p01): BACKUP_PORT=>BLOCKED_PORT 16:48:28:roletrns(B8322-p01): BLOCKED_PORT=>BACKUP_PORT 16:48:28:roletrns(B8322-p01): BACKUP_PORT=>BLOCKED_PORT 16:48:29:roletrns(B8322-p01): BLOCKED_PORT=>BACKUP_PORT 16:48:29:roletrns(B8322-p01): BACKUP_PORT=>BLOCKED_PORT 16:48:30:roletrns(B8322-p01): BLOCKED_PORT=>BACKUP_PORT 16:48:30:roletrns(B8322-p01): BACKUP_PORT=>BLOCKED_PORT 16:48:31:roletrns(B8322-p01): BLOCKED_PORT=>BACKUP_PORT 16:48:31:roletrns(B8322-p01): BACKUP_PORT=>BLOCKED_PORT 16:48:31:Age(B8322-p01) => Designated 16:48:31:roletrns(B8322-p01): BLOCKED_PORT=>DESIGNATED_PORT 16:48:31:roletrns(B8322-p01): DESIGNATED_PORT=>DESIGNATED_PROPOSE 16:48:31:roletrns(B8322-p01): DESIGNATED_PROPOSE=>DESIGNATED_PORT 16:48:31:roletrns(B8322-p01): DESIGNATED_PORT=>DESIGNATED_SYNCED 16:48:31:roletrns(B8322-p01): DESIGNATED_SYNCED=>DESIGNATED_PORT 16:48:31:topoch (B8322-p01): INACTIVE=>TCACTIVE 16:48:31:(B8322-p01) rx AGREEMENT flag ! 16:48:31:roletrns(B8322-p01): DESIGNATED_PORT=>DESIGNATED_LEARN 16:48:31:roletrns(B8322-p01): DESIGNATED_LEARN=>DESIGNATED_PORT 16:48:31:roletrns(B8322-p01): DESIGNATED_PORT=>DESIGNATED_FORWARD 16:48:31:roletrns(B8322-p01): DESIGNATED_FORWARD=>DESIGNATED_PORT 16:48:31:sttrans (B8322-p01): DISCARDING=>LEARNING 16:48:31:sttrans (B8322-p01): LEARNING=>FORWARDING 16:48:31:topoch (B8322-p01): TCACTIVE=>DETECTED 16:48:31:clearFDB (p01, B8322, other ports, 'DETECTED') 16:48:31:topoch (B8322-p01): DETECTED=>TCACTIVE When these 3 Bridges are connected "as a triangle", the problem is harder. I am working 2 days to analyze it, but may be you can help me ? The questions are: Is this a bug in my implementation, or flow in the algorithm, or appropriative behavior ? Do another implementations have the same problem ? Best regards, Alex |