A for Shawn, Randy, and others. (Yeah; the new emojis might get a little annoying in the wrong hands. Ahem.) This will be on the geeky side, Pamela, so maybe just ignore me and come back later if you decide to get deeper into the genetic weeds.
The "are your parents related" tool at GEDmatch that Randy mentioned is always a good first stop for anyone who's uploaded data to GEDmatch. What it checks for are what's called runs of homozygosity. Put really simply, all our bits of DNA come in one of four flavors, or letters (strictly, nucleic acids, but who's counting): A, C, G, and T. If you open up your raw data file in a text editor, you'll see two columns in each row (a row being a tested base pair, or SNP), and each of those columns contains one of the four letters. If the two letters in a given row are the same (e.g., AA, CC) then they're called homozygous, because they have the same pair of letters at that position. If you have a whole boatload in a contiguous sequence where two letters match each other at a given position (a locus), you have a "run."
GEDmatch doesn't have any extra options you can set for this utility; it's as simple as enter a kit number and go! And generally speaking, it's just fine as it is. If the two parents had fully identical DNA for long stretches, it will show up and indicate that the parents may share recent ancestors. If enough time (as in number of birth events) has transpired since those common ancestors, the tool won't indicate the parents were related.
It's important to know not just for endagomous populations, but also for families that might have relatively recent (or multiple instances of) pedigree collapse. This affects the average amounts of DNA sharing expected, and is why I always recommend it as the first stop for a new kit. For those who want to (or may need to, to get more information), the GEDmatch utility uses a technique developed by David Pike of Newfoundland. He has some free utilities available that allow you to get more granular, for example to set how many base pairs in a row must be identical to be considered a "run," and how many non-identical pairs are allowed before a run is considered broken and not really one run.
As others have noted, that 10.9cM segment isn't very big and, though it isn't within a known population pile-up region (segments along chromosomes that tend to be shared by very broad segments of a population because they aren't genealogically pertinent; they're passed down over many hundreds or even thousands of years), it is near the very beginning of chromosome 17 and is in an area that contains a number of protein-encoding genes...90 of them, in fact. In addition to broad population pile-up regions, we all have our own haplotype--or familial--pile-ups. Here's a recent article by Debbie Kennnett that explains it very well and includes some nifty visuals. Areas within protein-encoding genes--at least a lot of the important ones--don't mutate all that much. Understandably: rolling the mutation dice for important genes is risky. One example is an vital area of what's called the HLA complex on chromosome 6; it's involved in your autoimmune system. This particular area lives roughly between base pairs 25Mb (mega-base-pair notation) and 35Mb on Chr6. Have a look at the graph Debbie posted of her chromosome 6 and the frequency of matches as downloaded from her kit at Family Tree DNA. Can you spot the HLA region? Betcha did.
The point here being that you and your husband might have a ton of other matches in and around that segment on Chr17. There's a GEDmatch Tier 1 tool that let's you directly compare segments among all your matches, but as Vivian noted, you can accomplish mostly the same thing with the free "people who match one or both of 2 kits" option.
Last geeky wrench that I'll throw in the machinery: We tend to think of the segment sizes we see from GEDmatch and the testing companies as being a precise measurement. They aren't. Confusion rises especially when some reports show centiMorgan values down to the 1/100th. The centiMorgan ain't even a physical measurement at all; it's a linear equation interpolation of Base Pair A to Base Pair B, estimating the number of likely DNA crossover occurrences and using that to arrive at a guess about the relative genetic distance. <cough, cough> But there's a massive amount of mathematical estimates, probabilities, imputations, genotyping and, yes, educated guesswork that goes into the DNA results and predictions that we see. Even the start and stop base pair numbers you see from GEDmatch aren't precise (though they look like they are) because all that's tested are about 630,000 to 700,000 identified base pairs (SNPs) among the roughly 3.2 billion that make up your whole genome...or about 0.023%. So the test results can only look at the first and last tested base pair in a sequence, not the actual physical base pair numbers. Top it off, we call a segment a segment because we see, give or take a margin of error (ya know that bunching mismatch option in GEDmatch?), a sequence of identical SNPs...but there may be millions of base pairs within that sequence and we're only assuming that they're all the same because the thousand of them we actually tested were identical.
For the most part, it all works well enough for our genealogical purposes. But it pays to be informed, and to understand that segment sizes are not always what we assume them to be. (I'm flashing back to Diego Montoya in The Princess Bride: "I do not think that word means what you think it means.")
Another case in point is that what we have reported to us in cMs for segment size are what's called "sex-averaged" values. Not only do different versions of the human genome maps differ slightly from one another (and GEDmatch uses hg36 while almost all academic studies now use an iterative-level of hg38), but the male and female genomes differ significantly when it comes to the linear equations that estimate centiMorgans. The female genome experiences 30% or more crossovers than the male.
Let's take a look, using hg36 which is what GEDmatch uses, at that segment of yours on chromosome 17. Your computed value for it is 3.1cM; for your husband it's 17.7cM...ergo the averaged value you see of 10.9cM. When dealing with small sex-averaged segments, the actual gender-mapped value can completely zero out what was thought to be a segment of sufficient size. Once you're into the area of around 15cM, you're generally safe that a segment really is a segment. Randy talked about general reliability of small segments, so I won't other than to note an informal study done by a knowledgeable guy named Tim Janzen. He found that 7cM segments survived traditional trio phasing and proved to be valid only 42% of the time; 11cM segments proved valid 90% of the time; and 15cM segments 100% of the time.
I told you not to read this, and just to bookmark it and come back if you decide to really dig into genetic genealogy. You certainly may want to investigate that 10.9cM. I would, for fun if nothing else. But if there is pedigree collapse causing it, it's likely at least at the 4g-grandparent level; almost certainly not closer than 3g-grandparents. So it won't become readily apparent exactly which branch it's on without some significant digging. A popular, free app that I've come to love that can help with the effort is called DNA Painter. Other apps can do the same things, but DNA Painter is easy to use, flexible, you can share your results with others and, of course, free. The big plus is that using a chromosome painting tool like this to chase down that 10.9cM segment will mean you'll also be filling in matching segments for a bunch of cousins. The result is that you'll have most of the homework done when you discover the next new cousin.