Anyone who has a 6900?

[devnull]

We ordered a couple of 6900 in June but still did not receive any.
So does anyone has practical experiences? Is it stable? Are there any ceavats?

[devnull]

Received our first two 6900-40x going to test it in the labs.

First thought: Chassis looks good. Linux is clearly visible. Fans are pretty anoying at reload, but are fine afterwards (no load yet)
VRRP/OSPF needs Adavanced Licence - that sucks.. have to order a licence.. grml..


Going to do some basic test and make test a mc-lag

[devnull]

Some Updates more or less stable, we had some issues in the lab with VLANs being blocked on MC-LAGs. Which were gone after rebooting both 6900 ... that is not the thing i am looking forward for production.

Logfiles are very annoying: in default settings it fills the logs faster than you can see. After tuning logging, the switch is still showing errors, while being idle:

Code: Select all

Oct 18 13:10:36 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:199 L2Traverse memory freed- duplicated 2 times!
Oct 18 13:15:26 Core1 local0.info swlogd: bcmd rpcs info(5) _bcm_server_rpc_accept 473: Got Connection 56 my Socket 16- duplicated 2 times!
Oct 18 13:15:26 Core1 local0.info swlogd: slNi  info(5) aluClientCallbackHandler:196 L2_TRAV_CALLBAK_HNDLR_MSG_ID client deregistration!!- duplicated 2 times!
Oct 18 13:15:26 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:192 L2Traverse Disconnected- duplicated 2 times!
Oct 18 13:15:26 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:199 L2Traverse memory freed- duplicated 2 times!
Oct 18 13:20:16 Core1 local0.info swlogd: bcmd rpcs info(5) _bcm_server_rpc_accept 473: Got Connection 56 my Socket 16- duplicated 2 times!
Oct 18 13:20:16 Core1 local0.info swlogd: slNi  info(5) aluClientCallbackHandler:196 L2_TRAV_CALLBAK_HNDLR_MSG_ID client deregistration!!- duplicated 2 times!
Oct 18 13:20:16 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:192 L2Traverse Disconnected- duplicated 2 times!
Oct 18 13:20:16 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:199 L2Traverse memory freed- duplicated 2 times!
Oct 18 13:25:06 Core1 local0.info swlogd: bcmd rpcs info(5) _bcm_server_rpc_accept 473: Got Connection 56 my Socket 16- duplicated 2 times!
Oct 18 13:25:07 Core1 local0.info swlogd: slNi  info(5) aluClientCallbackHandler:196 L2_TRAV_CALLBAK_HNDLR_MSG_ID client deregistration!!- duplicated 2 times!
Oct 18 13:25:07 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:192 L2Traverse Disconnected- duplicated 2 times!
Oct 18 13:25:07 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:199 L2Traverse memory freed- duplicated 2 times!
Oct 18 13:29:57 Core1 local0.info swlogd: bcmd rpcs info(5) _bcm_server_rpc_accept 473: Got Connection 56 my Socket 16- duplicated 2 times!
Oct 18 13:29:57 Core1 local0.info swlogd: slNi  info(5) aluClientCallbackHandler:196 L2_TRAV_CALLBAK_HNDLR_MSG_ID client deregistration!!- duplicated 2 times!
Oct 18 13:29:57 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:192 L2Traverse Disconnected- duplicated 2 times!
Oct 18 13:29:57 Core1 local0.err swlogd: bcmd rpcs error(2) cbkHandlerDiscFn:199 L2Traverse memory freed- duplicated 2 times!
Oct 18 13:33:48 Core1 auth.info sshd[3943]: Accepted password for admin from 172.22.20.176 port 1487 ssh2
Oct 18 13:33:48 Core1 auth.info sshd[3947]: lastlog_filetype: Couldn't stat /var/log/lastlog: No such file or directory
Oct 18 13:33:48 Core1 auth.info sshd[3947]: lastlog_openseek: /var/log/lastlog is not a file or directory!
Oct 18 13:33:48 Core1 auth.info sshd[3947]: lastlog_filetype: Couldn't stat /var/log/lastlog: No such file or directory
Oct 18 13:33:48 Core1 auth.info sshd[3947]: lastlog_openseek: /var/log/lastlog is not a file or directory!

the sshd error shows it works with linux /you even have a busybox on it... But it probably must mature another release or two.

I am curious how they work in live environment.

[devnull]

Short update: I have multiple problems:
- with MC-Lags, which resulted in connectivity lost in parts of network - alcatel is still debugging the case, but they seem to have no Idea.
- with STP Root change traps from MC-Lag Peers, while STP is disabled on all ports (MC-Lags)
- with LACP working with a new 6850, but no lacp possible to a 6850 build end 2006

So i would have preferred to receive the boxes a bit later, after others found the errors

[larspo]

Stacking is not possible yet (even when told in the spec). This will be released in a firmware update april 2012 alowing it to stack 2 x 6900 (I was told). In a future release, there will be possible to stack 6 x 6900. There is a "workaround", but its not a true "stacking".

[skdvax]

Hi

looking for any updated info on stability and experience with more recent firmwares

thanks,

[benny]

It has been a year, the software improved a lot. AOS 7.3.1.R01 is available that also features SPB (Shortest Path Bridging) and many Data Center related features.
From my experience pretty cool and stable box, waiting for the final version of the Virtual Chassis to test.

Benny

[devnull]

I had some strange experiences with 7.3.1 on OS10k, and as i don't need SPB now (i don't have enough 6900 to gain advantage compared to MC-Lag) i prefer to wait before upgrading to 7.3.1.

I'm happy with MC-Lag running 6900 now. Most of my former problems were because of a mixed cisco/alcatel network.

[skdvax]

Hi Guys.

thanks for your comments.
devnull - can you expand on your comment "problems were because of a mixed cisco/alcatel network"?

thanks,

[devnull]

- configure a compatible STP on MC-Lag, ALU and Cisco i went for 1x1 pvst+ (compatibility).
- Configure that on 6900 as well even while they should not do STP. (They will only filter STP Packets on the MC-Lag Ports when set to the "right" STP)
- Avoid VTP, DTP. They will do strange things when connected to ALU Switches
- Use LACP, while mixing etherchannel and static agg may work lacp is far more predictable. Use it!
- Don't expect much help from support when you explain that you have a mixed network. It's allways "other vendor"s fault.