Unsolved
This post is more than 5 years old
9 Posts
0
37489
July 20th, 2012 12:00
PowerConnect 6248 High Ping Times when LAG Enabled PLEASE HELP
I have two 6248 switches connected with a 10Gb Fiber link. I did this because of the distance they are from each other. No problems here, everything works great between the two switches.
The issue I am having is that on the main switch I have each of my servers in a LAG to get 2 Gbps link speed out of each one of them. Still no problem here it seems to work okay.
When i ping the server that are in the LAG I get <1 ms reply... The issue is that when i ping the switch, I get higher and higher ping times (regardless of which workstation I ping from) and a high TTL.
Pinging 10.1.1.254 with 32 bytes of data:
Reply from 10.1.1.254: bytes=32 time=6ms TTL=64
Reply from 10.1.1.254: bytes=32 time=62ms TTL=245
Reply from 10.1.1.254: bytes=32 time=106ms TTL=245
Reply from 10.1.1.254: bytes=32 time=34ms TTL=245
The second switch that is is not using any LAGs has a totally normal ping of <1ms and a TTL of 64.
To make issues worse, I can login to the web console of the second switch with no problems whatsoever. The switch that is having high ping times just times out or says connection reset when trying to login to the web console. What confuses the hell out of me is that when i disconnect all of the servers which are using the LAGs, the ping times return to normal and I am able to login to the web console.
What did I do wrong?
Finally, I would assume that because I have no loops in the network at all, that i do not need to have STP enabled...
Can anyone please help me? Oh yeah, I don't have any VLANs or anything special going on other than the LAGs for the switches and the 10 GB fiber between the two switches...
Thank you,
Ryan
DELL-Willy M
802 Posts
0
July 20th, 2012 12:00
Are you configuring the LAG as static or dynamic?
console(config-if)# channel-group 2 mode auto (LACP/dynamic)
console(config-if)# channel-group 2 mode on (Static)
How is your team set up on the servers? Are you matching the LACP configuration on the switch?
I would also recommend that you have the latest firmware installed.
v3.3.3.3
www.dell.com/.../DriverFileFormats
Hope this helps,
Keep us updated if you can.
ryanpair
9 Posts
0
July 20th, 2012 13:00
I will try this in the next few days. Thank you for the suggestion. I will post an update after I attempt this.
Thank you,
Ryan
ryanpair
9 Posts
0
July 23rd, 2012 14:00
So i checked the config, the switch is setup using LACP (dynamic) and the server is setup using the Broadcom utility as an 802.3ad (Link Aggregation). I did both of these via the GUI in both the switch and the BACS utility. As far as i can see they are setup the same...
I am updating the firmware today in 1 hour when i can take the servers down.
ryanpair
9 Posts
0
July 23rd, 2012 16:00
I tried the following...
Updated firmware
Took all servers that are in the LAG out of the equation. When no servers are connected, ping times are <1 ms and the switch web server is accessible.
Tried changing to a regular LAG (not LACP) on both switch and main server. Plugged it back in copied config to boot and rebooted switch. Things are fine for a few minutes, then high ping times, high TTL, and switch web server unresponsive.
Changed back to LACP, both switch and server, same problem as above.
Turned off STP (i only have two switches and no room for loops in the network)
None of the this has helped... Any Ideas? Strange how when i turn off the LAG or just leave the LAGs disconnected how everything is fine...
DELL-Willy M
802 Posts
1
July 23rd, 2012 17:00
You might need to update the BACS driver and firmware. Also, try setting the port-channel to portfast.
spanning-tree portfast
Use the spanning-tree portfast command in Interface Configuration mode to
enable PortFast mode. In PortFast mode, the interface is immediately put
into the forwarding state upon linkup, without waiting for the timer to expire.
To disable PortFast mode, use the no form of this command.
An interface with PortFast mode enabled is moved directly to the spanning
tree forwarding state when linkup occurs without waiting the standard
forward-time delay.
Example
The following example enables PortFast on 1/g5.
console(config)#interface ethernet 1/g5
console(config-if-1/g5)#spanning-tree portfast
ryanpair
9 Posts
0
July 24th, 2012 06:00
I checked and i have the latest BACS software, driver, and firmware. I will try the STP portfast suggestion next. It will be a few days until i can shutdown the network again to test. I will let you know what I find.
ryanpair
9 Posts
0
July 24th, 2012 06:00
Interesting observation, I disabled all of the LAGs on both the servers and the switch. everything is hooked up without a LAG now and the switch responds to everything as it should. (low ping times, good ttl, and webserver responds)
richter_13
18 Posts
0
July 24th, 2012 08:00
Just a quick note, picking up from something you have mentioned in your posts.
how are the VLANs setup on your switch ? are you using the default VLAN for production.
the reason i ask is because of your statement that the switch management itself becomes unresponsive.
in VLAN 1, the default VLAN. your switch behaves like a generic IP host, so any issues with Broadcasts, loops, storms etc on this VLAN directly impacts the switch CPU. one of the results would be the management access becoming unresponsive. thus, in deployments, i frequently orphan VLAN 1 completely. use Out of band if available. if not, use a separate inband VLAN for management.
secondly, have a look at the switch logs. if the reason i have cited above is the cause, you will see a lot of activity. perhaps STP related.
since it is localized on the second switch, we are likely to find some clue in the logs.
one thing i have at times observed with Powerconnect switches is that while portfast transitions port to forwarding more quickly, (as par standards), it does not always suppress TCNs and BPDUs as i would expect it to. this behaviour does change firmware to firmware.
do post back if you see anything interesting in logs.
ryanpair
9 Posts
0
July 24th, 2012 09:00
I'm not using a VLAN. Should i remove the "default". I dont have a need for a VLAN, we are using a basic network setup with no VOIP phones or need for separation.
richter_13
18 Posts
0
July 24th, 2012 09:00
This means you are using the default VLAN, just as i had predicted.
with the switches, even with no VLAN configured, all ports would remain part of the default VLAN 1 which cannot be disabled and is always present.
check it out via:
show VLAN
you will see port membership of all your ports in VLAN 1.
**
My train journey is coming to an end. :)
i might revisit this thread later to see if any further advice from my side would be of assistance.
cheers