A RCI Glitch is Apparently Back

[jlwquilter Senior Member Find all posts View Profile Close jlwquilter]

I had to call RCI this morning. I wanted to exchange into a week that was showing on the website as available but not allowing me to grab it.

I ended up talking to a supervisor (very nice and seemingly knew what she was doing) about the week. She used the phrase "the week is kind of floating". When I asked what that meant, she said that the week was actually on hold by someone and not truly available right now.

So... we may be getting sightings of weeks that are hold and not really available (phantom weeks??), which result in the all too often seen screen of 'transaction can not be processed at this time. Call RCI' when we try to exchange into them.

I believe Lawren has reported this kind of glitch after other RCI updates, enhancements, etc.

[BoardGirl Senior Member Find all posts View Profile Close BoardGirl]

That's annoying. If it's not available, remove it from the availability list!

[jackio Gold Contributor Find all posts View Profile Close jackio]

I'd put in an ongoing request so that you can grab it if the person releases it.

[bnoble Senior Member Find all posts View Profile Close bnoble]

If it's not available, remove it from the availability list!

Reading between the lines, and combining with some reported conversations with Tech Support, it appears that some "fixes" they added to improve the response time problems from early/mid-June relaxed the "consistency" between the front end (the web site) and the back end (the legacy system that actually keeps track of inventory really-for-real). So, at any point, the web site may be slightly out of date with the back end, and this isn't determined until the back-end is contacted to remove the unit from inventory.

It might not be possible to fix this problem without either (a) re-introducing the hideous daytime performance from early June or (b) re-architecting the entire front end.

[BoardGirl Senior Member Find all posts View Profile Close BoardGirl]

Probably it would be the middleware and not require re-architecting of the front end. If they are using an intermediary database that refreshes rarely, that's the problem. And that's what it sounds like - original search goes to intermediary db and shows available units known to that db per last requery but fetching one of those units goes back to mainframe. that's my guess, anyway. Explains performance increases because searches are "local" vs round-trip from mainframe until you want one of those units, then, OOOPS, only existed in the other database, error, etc etc.

Serving a website from a mainframe being also used internally presents problems with data consistency/refresh (which is why I think they have a second "web database"). But, these are not new problems.

[djyamyam Senior Member Find all posts View Profile Close djyamyam]

So... we may be getting sightings of weeks that are hold and not really available (phantom weeks??), which result in the all too often seen screen of 'transaction can not be processed at this time. Call RCI' when we try to exchange into them.

A memo went out to all the VGs so you'll notice that they all the term "stale dated" now in these types of situations. Love the food reference term but it's still not there any way you cut it

[bnoble Senior Member Find all posts View Profile Close bnoble]

Probably it would be the middleware and not require re-architecting of the front end.

Yes, that's a more precise way of stating it. I tend to think of the middleware as either one or the other, but that's not really correct.

"Stale" is a term often used in cache designs, actually.

[lawren2 Malibu Beach Barbie & A Super Moderator Find all posts View Profile Close lawren2]

Probably it would be the middleware and not require re-architecting of the front end. If they are using an intermediary database that refreshes rarely, that's the problem. And that's what it sounds like - original search goes to intermediary db and shows available units known to that db per last requery but fetching one of those units goes back to mainframe. that's my guess, anyway. Explains performance increases because searches are "local" vs round-trip from mainframe until you want one of those units, then, OOOPS, only existed in the other database, error, etc etc.

Serving a website from a mainframe being also used internally presents problems with data consistency/refresh (which is why I think they have a second "web database"). But, these are not new problems.

They must have as VGs can see availability when the web on-line is down.

and no these are not new problems. The same thing happened with phantoms when they first introduced this new way of searching for exchanges.

They were able to fix it then and they probably will be able to fix it now but it won't be pretty while they do it.

[djyamyam Senior Member Find all posts View Profile Close djyamyam]

"Stale" is a term often used in cache designs, actually.

You learn something new every day. Thanks for the clarification

[bnoble Senior Member Find all posts View Profile Close bnoble]

And, as "old" inventory is removed "late", that probably means that "new" inventory is added "late" too.

It would be worth considering initiating an ongoing search for anything you've really got your heart set on, just in case those match first.

[jpaviles Member Find all posts View Profile Close jpaviles]

Are they any IT folks out here?

Does anyone know the full architecture of the site/database?

I see it mentioned they have a mainframe backend DB.. anyone know what the front end is coded in?

Doesn't appear to be anything Microsoft so I'm guessing it's a java platform.

I'd love to create their tech design documents for them and do the QA testing!

[jlwquilter Senior Member Find all posts View Profile Close jlwquilter]

I'd put in an ongoing request so that you can grab it if the person releases it.

Got it covered! With a little help from a friend

[jlwquilter Senior Member Find all posts View Profile Close jlwquilter]

I always figured there was some type of "batch" running going on at RCI.

That would explain how ongoing searches are "missed" and matches show up online. The ongoing search program is not always (ever??) run at the same time new inventory is added.

[ebram Senior Member Find all posts View Profile Close ebram]

Nothing wrong. It's like when they put "sale pending: sign over a for sale real estate sign. Or if someone reserves an opera ticket and has not cleared his CC yet.