01-21-2003 08:03 AM - edited 03-02-2019 04:23 AM
Hi
Exact message:
2003 Jan 21 16:46:51 MET +01:00 %PAGP-5-PORTFROMSTP:Port 2/20 left bridge port 2/20
2003 Jan 21 16:47:10 MET +01:00 %PAGP-5-PORTTOSTP:Port 2/20 joined bridge port 2
/20
I know the URL http://www.cisco.com/warp/public/473/34.shtml#PAGP_MESSAGES
I dont think there are
Speed/duplex mismatch
Faulty cable
Faulty NIC or other end station problem
Faulty switch port
Other misconfiguration
The intervall between leaving and joining is always 17 to 20 secs. The user does not notice anything, I can ping the connected stations all the time. I have this on many ports of the Cat2948 switches; there are no alignment errors.
I have the same problem on Cat6513, SW 6.3(8)
What shall I do ?
Thank you very much
Alfred Neumann
01-21-2003 08:39 AM
01-29-2003 01:26 AM
ok. Thank you very much.
Alfred Neumann
02-06-2003 12:22 AM
Did this document solved your problem because we have the same problem on our Catalyst 6509 switches. On one switch we have interuptions of 3 seconds and on the other switch interuptions of 17seconds.
We have the same problems on Catalyst 3500XL switches.
These are messages that we get in the log file of the cat 3548XL. Debugging of the ethernet controller and interface was on.
.Feb 5 15:17:38: %LINK-3-UPDOWN: Interface FastEthernet0/23, changed state to d
own
.Feb 5 15:17:39: PHY CHANGE: port 25, 0x0000 to 0x0031
.Feb 5 15:17:39: PHY CHANGE: port 25, 0x0031 to 0x0000
.Feb 5 15:17:42: %LINK-3-UPDOWN: Interface FastEthernet0/23, changed state to u
p
02-18-2003 11:53 PM
Generally we have the 17 sec problem on Cat6509 SW 5.5(14), moduls WS-X6248A-TEL, on Cat6513 SW 6.3(8), moduls WS-X6548-RJ-21and on Cat2948G. The power-management feature on some NICs etc. have no impact .
If the switchport is in autonegotiating mode, fix to 100 and fdx will help. But we have a lot of HP-systems with fixed 100 and fdx ports, which have the 17 sec problem as well. My uncertified feeling right now: it depends also on utilization ??? more load, more 17 sec left bridge/joined bridge ???
Some feedback ?
Fred
02-19-2003 06:05 AM
I had the same problem on a few servers here. I was forced to use 10/half which solved the problem.
I believe this to be due to the poetic license some venders use while writing their 100/full driver. Most networking gear adheres very strictly to the RFC while some nic venders are liberal. If you search on the cco you will find documentation supporting this.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide