«

»

Oct 23

The beatings will continue…

…Until morale improves.

Trying to run 4, FCIP trunks over a half a DS3 is a lot like raising a teenager.

Sometimes it looks like it’s working, but in reality it’s just screwing around playing video games.

Actually, my favorite is that “Raising a teenager is like trying to nail JELL-O to a tree”  I’m feeling about the same level of frustration.

What’s basically happening is that the link is fine, as long as we’re not doing anything silly like, oh, PASSING DATA over it.  THe minute we start moving data the link gives up and goes to Palm Beach for the holiday.

I tried to explain to both EMC and the customer at the start of this engagement that replicating four Symms over even a full DS3 is very…optimistic.

So I’ve spent the last three days solid beating my head over this, more than 18 hours a day (except for yesterday which involved 7 hours + 8 hours travel time.

6 comments

Skip to comment form

  1. Jesse

    Ok, it helps if the MTU settings on the cisco match the MTU settings on the core-network switch.

    The MTU settings on a 9216i default to 2300, I’m assuming to accomidate the 2112byte fibrechannel packet.

    Standard MTU settings for ethernet is 1500… If one switch is set to 2300 and the other is set to 1500, one ends up overrunning the other one with data.

    Hence, until we actually started pushing real payloads, the link appeared fine.

    The Best solution here would have been to set the core switch to 2300, to accomidate the 9216. However I wasn’t willing to take the chance that everything else needed to be changed as well.

    So we set 9216i’s to 1500 and it worked.

    76 hours last week, not counting the flight home on Sunday (following week). But time flies when you’re having fun, right?

  2. Heh, were still trying to get our SRDF working….. Plugged in the RA’s to the Cisco and no lights!!!! Nothing in the flogi database. We think its a setting on the cisco, the old links were using CNT’s and we are going from the CNT’s to the cisco. The FCIP TRUNK was created with no problem after of course remembering to add the route. We have a case open now trying to figure it out. They were talking bin file changes, etc etc… but that is gonna be a last resort.

  3. Jesse

    What kind of RA’s are you using? If you’re using converted FA’s, (Called RFA’s) it should work, though if you were going to CNT boxes you’re going to have to change the Point-To-Point flag on the directors

    A CNT doesn’t look like a switch to an FA, it’s FCAL. So if your FA is in AL mode you aren’t going to see a login until you change this.

    I believe there is a way to run this from the command line using configuration manager, but I’ll have to research the exact syntax for you.

    The easiest way is to call EMC and tell them you need RFA8a, and RFA9a converted from AL to PTP. Though depending your support contract it might cost you. 🙂

  4. SG, don’t feel so bad — I had the same MTU problem with Cisco 9509’s and FCIP as well. Too bad I didn’t read through this post until now, otherwise I would have said something 😉 I spent several days on it as well.

    zombielun — Strangely enough, the issue you described sounds exactly like a case I just assisted with on Friday morning (shortly after you posted the comment above).. Resolution was to replace the ESCON RA’s with Fibre Channel RF’s. I’m guessing that was probably your case..

  5. Jesse

    The part that drove me nuts was that it seemed to be working so well, until we actually started moving data and then it came crashing down without any real error. Just a link-drop and the Symm automatically suspends the SRDF session but no real errors.

    I’ve seen this happen before, had a case recently where the TC (Pre-Sales Technical Consultant in EMC speak) ordered almost *ALL* the wrong parts for a data-migration. We ended up scrambling at the last minute to find older (almost antique) parts for 7-8 year old Symms.

    What a pain that was. 🙂

  6. SANiT

    SG.. U R Hilarious. It’s good to have a sense of humor when working with this junk.
    Been Doing it for years and it seems to be about an 8 to 1 ratio of Major Pain in the a$$ projects to Oh? That was easy types of projects.

    Hang in there. Good job.
    P.S. in EMC speak TC actually means “Technical Complications” brought about by injecting yet another layer of sales/marketing people who have no idea how any of this works. (IMHO)

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>