FOS 8.2.1 – the one that is required.. or not?

Like clockwork Brocade releases new FOS version around every 6 months. No news here. FOS 8.2.1 is however a release you may need to pay special attention to especially if you have X6 director class switches hooked up to a 240 volt, 50 Hz power mains as well as sitting between a rock and a hard place with the 7800 extension switch but don;t have the budget to go to a relatively pricey 7840. One other thing is the change in licensing hardening on pizza-box switches which makes the upgrade to this release a one-way street without being able to go back.

Ok, so whats this deal then wit the X6 power. I’ll get to that a bit further up.

7810 extension switch

First the support for a new extension switch. The 7810 is a new switch which has beefed up capabilities of the 7800, is architecturally different but provides similar functionality and interoperability as/with the 7840. This means that it can also be connected to the 7840 or SX-6 extension blade which makes it ideal for branch offices with remote replication requirements without having to budget for a full-fledged 7840.

The Brocade 7800 has had an impressive run but has come to an end w.r.t. options required in today’s datacentre’s. The speed limitation on ethernet-ports as well as the maximum of 8G on the FC ports make it hard to utilize the grunt of modern storage arrays and IP connectivity requirements if IPExtention is used. The other limitation of not being able to set up FCIP tunnels with newer equipment was also a a thorn in many customers eyes if they wanted to upgrade their core-DC infrastructure but still had to rely on the FX8-24 or another 7800 for FCIP connectivity. The 7810 resolves that and is quite comparable to the existing 7840.

It comes in two flavours, licensed and unlicensed. The difference is mainly by having a limitation on the number of available ports, speeds and features. There is one upgrade license and for the price-point you should contact your sales-rep. Functionality wise I would definitely recommend to purchase it but hey, I’m not the one who has to pay. 🙂

There is also a rather large bucket-list of functions and feature that is not, and will not, be supported. Ficon, Hot-code-load, FCoE and Virtual Fabrics are not available.

Optics

This version also has support for a new SFP which allows for long-distance FC connections to be made on 32Gb/s up to 25KM. In previous releases you had no other option that to use an intermediate switch in between data-centres if the distance was over 10KM and you had a 32Gb/s bandwidth requirement. This SFP will bridge that gap.

Counterfeit protection

As there are always crooks in the world who want to make money on what other people work hard for, the IT industry is not immune either. As it turns out some of these people line-up an array of switches they buy on e-bay for a buck-and-a-half (as a figure of speech) without any licenses and simply try to brute-force the generation of a license-key onto that switch after which they sell it with licenses and a significantly beefed up price-point obviously. You’ll see this most often in the lower priced switches like the 6505 and G610. The director class switches are rarely seen on these online shopping sites but are also not 100% immune to this criminal behaviour.

To counteract that Brocade has implemented a new encryption algorithm as well as the demand for much longer  license keys in FOS 8.2.1.There is now also a limitation on how often you can try a “licenseadd” command within a certain time-period before the switch bricks itself and will be permanently disabled. And no, E-bay won’t re-reimburse you on that. 🙂

Here is the catch.

Already existing installations of these platforms upgrading to FOS 8.2.1 cannot be downgraded. 

RESTAPI

You may have read my previous article on 8.2.0 where Brocade introduced the RESTAPI. In this version the API is extended with a fair few modules which allow you to monitor/manage the switch with your own tools without having to rely on basic shell-scripting via the CLI. This has been a long time wish for me as I’ve seen many switches being scripted against to their ears and get crippled to the extent they eventually trip over. The RESTAPI can prevent against that plus if provides a much more structured methodology in obtaining information as well as incorporating it in enterprise monitoring and management tools.

R_RDY on XISL’s

In previous releases the XISL (basically the virtual ISL connecting two base-VF switches) worked only in VC mode 1 effectively creating 4 data VC’s to be used to disperse traffic over these virtual channels. As I have written before this requires any third party equipment (most notably DWDM/CWDM kit) to support this if they use transponders instead of optical switching. With this option the ISL can be set into R_RDY mode even for XISL’s wich would flick the ISL into a single channel which should be supported by all 3rd party equipment. If it doesn’t work after this some other parameters may need to be adjusted.

Management Interface Rate Limiting

This is a setting done via the “configure” command and enabled rate limiting on the ethernet management interface to prevent DOS attacks. I haven’t been able to figure out what it does exactly under the covers. At first I thought some sort iptables rule which would drop packets based on certain criteria but that doesn’t seem to be the case. Then I have a look at the ethernet qdisc values but these also seem to be the same irrespective if this setting being enabled or disabled. Might need to spend some time on this.

There are a few other bits&bobs that have been added and changed but not really earth shocking.

As it happens last week Brocade released its first patch cycle on the 8.2.1 version which obviously would still be going through the OEM testing cycles. As far as I’m concerned at this stage (I loaded this version on a lab X6 system) I have not seen an issue so far and the 8.2.1a version should be the proper one for all Gen 5 and Gen 6 platforms at the time of this writing. (but that being said my lab is not that extensive and I’ll leave the testing up to the test labs for official qualification)

The power problem

A somewhat nasty issue if you have a fully loaded X6-8 or X6-4 and you have a non-redundant power-supply configuration (either not installed or temporarily missing mains power to one or more PS). A defect has been identified in the PS firmware (yes NOTHING runs without software these days) where an incorrect RMS (Root Mean Square) is calculated resulting in the fact that the PS will bounce if multiple sequences happen in a short time. If you have an electrical background you’ll know that the RMS is used to align the AC sine values to the DC required reference voltage. If something goes wrong here and the calculated AC voltage is deemed too low for DC conversion some thing needs to give.

This may then result in blades being powered down to the extent the required DC voltage use is at or below the point the AC RMS voltage can provide that load. It turns out this only happens when the demand on the DC side is high and the provided AC side is low, i.e. a non-redundant power supply configuration. In addition to that it also only happens on power-supplies providing 240volt and running on a 50Hz grid frequency. Presumably the hardware engineers in the US had some difficulty engineering and testing this on 240/50 mains.

Anyway, the issue is resolved with this release by the use of a new cli tool which will upgrade the power-supplies to the latest fixed version. Brocade could have done the upgrade as part of a normal FOS upgrade however as it does not know the impact of a switch in your environment it needs some planning as, as mentioned, a switch may go down if not enough power-supplies are installed in the switch.

You might be familiar with the output of psshow in earlier FOS levels showing some basic information about the power-supplies:

psshow -v:

Power Supply #1 is OK
V10L03, AGC2M03L7A1 ,23-0000067-01, C,DELTA,ECD14020006 ,06,AGC2M03L
Power Supply #2 is OK
V10L03, AGC2M03L7A9 ,23-0000067-01, C,DELTA,ECD14020006 ,06,AGC2M03L

The new output shows much more information:

psshow -v:

Power Supply #1 is OK Temperature is 22.00 C
V10M26, DUC2M26M29Y ,23-0000161-01,A0,DELTA,ECD16020042 ,00,DUC2M26M
Primary FW Version: 3.1
Sec LLC FW Version: 3.1
Sec COM FW Version: 3.0
Voltage input: 236.50 V Voltage output: 49.61 V
Current input: 1.69 A Current output: 7.12 A
Power input: 388.00 W Power output: 352.00 W
Fan1 Speed: 8384.00 RPM Fan2 Speed: 8704.00 RPM
Inlet Air Temp: 22 (C), 71 (F) 
PFC Heat Sink: 18 (C), 64 (F) 
LLC Heat Sink: 18 (C), 64 (F)

Power Supply #2 is OK Temperature is 22.00 C
V10M12, DUC2M12M1HG ,23-0000161-01,A0,DELTA,ECD16020042 ,00,DUC2M12M
Primary FW Version: 3.0
Sec LLC FW Version: 3.0
Sec COM FW Version: 3.0
Voltage input: 236.00 V Voltage output: 49.33 V
Current input: 2.00 A Current output: 8.38 A
Power input: 460.00 W Power output: 412.00 W
Fan1 Speed: 8640.00 RPM Fan2 Speed: 8832.00 RPM
Inlet Air Temp: 22 (C), 71 (F) 
PFC Heat Sink: 18 (C), 64 (F) 
LLC Heat Sink: 20 (C), 68 (F)

The interesting part is obviously the PSU FW versions. After the upgrade they should show:

PRI - 3.4
LLC - 3.2
COM - 3.1

This would indication that you would not run into surprises regarding power with blades being turned off. Once more check with your vendor which version are tested/supported.

Regards,

Erwin

Print Friendly, PDF & Email

About Erwin van Londen

Master Technical Analyst at Hitachi Data Systems
Brocade Technical, Troubleshooting , , ,