Newsletter November 2008
Hi All,
Alwin here with your Collectorz.com newsletter.
I know how you love the "Meet the Collector" stories, but this time I thought I'd give you a Collectorz.com behind-the-scenes story instead... ... one about how we ROYALLY messed up.
This should be loads of fun!
But first, a tip about a topic we get lots of questions about:
TIP: How to setup your CueCat scanner
======================================
Many Collectorz.com users own a CueCat scanner. And it's no surprise, this
handy little scanner makes cataloging about as easy as breathing...
But we are getting the same question over and over again:
How do I setup my CueCat to work with my Collectorz.com software?
So here it is, a complete step-by-step guide on how to do that:
Step 1: Plug its USB connector into a free USB port.
And that's it! Just plug it in and start scanning barcodes!
Our software has integrated support for the CueCat, so there's no need to install drivers or anything. Also, you don't need to use the Initialize or Find Scanner commands in our software, those are for Flic and Opticon laser scanners.
The CueCat will work straight out-of-the-box! That's the beauty of it!
How we ROYALLY messed up...
=============================
Remember how we unveiled Book Collector 6 last month? Well, we've had some fun with that one :-)
The 6.0 release was released on time and was very successful too. Thousands of Book Collector users upgraded to BC6 in the first week!
But then the problems began: we started receiving emails complaining about searches being slow, sometimes taking up to a minute to complete. It looked like our database server wasn't able to cope with all the new searches.
Before the BC6 release, it was running fine, handling four media databases (for CDs, DVDs, comics and games) without problems. But the addition of the book database system somehow tipped it over the edge.
Of course, Sven, Dinh and I got to work right away, rewriting SQL queries, optimizing our indexes, etc... Which helped ... a bit .... for a short while ... But as more people upgraded and new users started buying v6, the problems got worse. It was ugly, really.
During peak hours nothing worked, the entire server just grinded to a halt. We were monitoring the server all day, resetting it once in a while for some temporary relief. And since all databases run on one server, the other programs begin to exhibit slow searches too. Music collector was the worst. Ouch...
You can imagine what happened to our support inbox... it practically became a full-time job answering all those support tickets.
It soon became clear that the server hardware was not sufficient to handle all five media databases. A hardware upgrade would be the only way to solve this once and for all.
But that's easier said than done. Luckily our friends at pair.com were very helpful, but still, it would take SIX days before they could have our new server up and running....
In the meantime Ronald and Sven were examining some reports from users about results that were, let's say, "less than optimal": books not being found while they were clearly present on Amazon. Some even reported they were not getting results AT ALL, never ever...
At first, we were just scratching our heads, not understanding what was going on. Was this a side effect of the dying server? Or were we looking at a completely different problem, or maybe multiple problems? Hard to tell at that point.
Luckily, after a couple of detailed reports from some helpful users, we found one problem: our system wasn't handling ISBN-13 results correctly. That is, it was treating them as different from their ISBN-10 counterparts whereas they of course were the same book.
Nothing a day of coding couldn't fix ... so we did just that... we fixed it up lightning fast.
It didn't explain the "no results at all" problems, but I'll get back to that later...
On Friday November 14 we could finally switch to the new server...
(for geeks: a Core 2 Quad Q6600 2.4Ghz CPU, 1066MHz Front Side Bus, 8 GB
RAM, running FreeBSD 64-bit, PHP5 & MySQL5)
... and it's FAST! All searches were returning results IMMEDIATELY ...
Phew...
With the performance problems gone we began to have to a clearer view of the remaining "no results at all" problems. Ronald and I were getting the feeling that this wasn't a server problem at all, maybe it was some stupid little bug in the program itself?
That Saturday one smart BC user from Belgium, Patrick Bout, gave us the final clue: he WAS getting results when we started a NEW database, but not while his own database was open. Huh??
Patrick had attached his own database, which helped us reproduce and then fix the problem. Ultimately it was a bug in the code that checks the existing database to see if a book is already present. Patrick is the man!
On Monday, 18 days after the 6.0 launch, we released Book Collector 6 build 3, fixing this bug and several other smaller problems.
Yep, I admit it: We messed up BIG TIME here.
We should have foreseen the server problems, we should have caught those
bugs before the release.
On top of that, it took far too long to fix all problems...
... not our usual turnaround time ... far from that ...
So I would like to apologize to everyone who purchased BC6 in the first few
weeks after the release.
And of course, a big THANK YOU! to everyone who helped us by reporting the
problems, both by email or on the forum, great work!
Anyway, the server is spinning happily now and BC6 build 3 has proven to be
nice and stable.
Book Collector 6 really rocks now. No really... it does ... now ...
Still using an older version of Book Collector?
Then now's the time to get your upgrade:
Book Collector Upgrade Page
Don't own Book Collector yet?
Then don't forget about our "PRO for Standard price" offer:
Book Collector PRO for Standard price offer Page (valid until December 1, so hurry)
Enjoy,
Alwin Hoogerdijk
President, Collectorz.com