05-08-2019 12:33 AM
I have eight Brocade 6510 SAN switches, each four are stacked. Stachs are isolated. I ran the SAN health audit, and got two reports, one for each stack. Now, I have concern about the following and I need you please to clarify and tell me what should I be doing about this.
05-08-2019 01:41 AM
I'm working on optimizing our SAN fabric, I've found tons of latency warnings on the ISL ports, and I've also realized that the BB credit on ISL ports is FC: 8 only, I think it should be higher thatn this. How could I increase it, and waht would you recommend for me please?
I'm not an advanced expert with these thjings, is there a good training course that I can take?
05-08-2019 10:23 AM
I merged both Thread in the single one, since handle the same issue/question.
here in the community are a lot of CRC Thread and Buffer Credit.
the Buffer credit value is 8 by default.
about the issue with the latency,
with the command sfpshow -operands ( see command reference manuals )
check if you have such SFP begin with follow Serial Number
05-09-2019 01:39 AM - edited 05-09-2019 01:44 AM
We have a strachetd datacenter, the distance betwen both is less than a KM, connected with dark fiber cables. I have two Brocade 6510 switch stacks, four on each stack. Each stack is spreaded accross the two sites, two here, and two there. I attached a diagram for the current connectivity. There's only a single ISL cable between both sites, should't it be a pair of cables between sites for redundancy and load balancing? as we are using MetroClustered SA storages, and servers from both sites access storage on the other site.
05-09-2019 08:23 AM
What is the topology of the fabrc? (Hub-Spoke/Star? Hybrid? Mesh?)
Is the issue seen in both fabrics?
Do any of the SFPs autonegotiate to a lower speed?
sfpshow -all | grep -e Port -e Power (Case sensitive)
This will show the RX and TX Power for all SFPs. Look for a low RX Power as compared to the TX Power. If you see this, you might have a cable issue. (Low meaning double digit uWatts)
Wait for an hour or so, then
This will show the stats from the time of the last stats clear. If clean, the issue may have been seen during fabric forming/power on of servers, etc. Continue to monitor. If not clean, it should be a little bit easier to pinpoint the issue.
There are a couple of free classes available here that can help you out. I highly recommend reviewing them.
This isn't a be-all list. But it is a good start.
05-09-2019 08:27 AM
The short answer is "yes".
Pose the question to those in charge as something like this: "What happens if [port connecting the two buildings] goes down? This could be switch port failure, an SFP failure, a cabling failure, etc...
05-10-2019 05:39 AM
Thanks for the valuable information. I will go through it and let you know the result. I joined this comlex environment couple of months ago, and my experience with Brocade is just with single/standalone swithces, not with complex fabrix, and I'm learning about this architecture from Brocade tutorials now, so, thanks for being patient with me till I understand everything to be able to better optimize our equipment :).
Answering your questions here:
What is the topology of the fabrc? (Hub-Spoke/Star? Hybrid? Mesh?) Not sure, attached a diagram for one of the fabrics, and both are identical.
- Is the issue seen in both fabrics? yes on both fabrics, only in the local ISL per site, I mean, not on the ISL of the inter-site connection.
- Do any of the SFPs autonegotiate to a lower speed? Yes, All are set to AutoNegotiate, and I've found that one of the ISLs detects the speed as 8G, despite the fact that the SFP is 16G. I will set the speed to manual.
The weekend is here now, so, I will run the comands you've mentioned by next week adn let you know the resukts, meanwhile, resoind my feedback for now if you'd like to.
05-10-2019 05:41 AM