Reaching Mars with resupply-levels of material (not just a crate of potatoes) is on the absolute vanguard of current tech, and given the laws of physics and economics not experiencing a giant upheaval, will stay there pretty much indefinitely.
A lot of specific materials, parts, people and facilities are absolutely neccessary to make a resupply run to Mars viable - there is not that many spaceports, rocket factories, hardened chips, ... that it would be inconceivable, or even unlikely, to see the accidental death of a few people, an unrelated crisis pinching some distant supply line, or a mild catasrophe wrecking some important infrastructure set back a space programme for years.
The supply rocket fails spectacularly, with no readily apparent cause - a billion dollars down the drain. (Emergency-) Supplies on Mars will last another three years, so the consensus is to thoroughly investigate the failure, revise protocols, and launch at the next window in 2 1/2 years.
Turns out the failure was due a Y2K-like thing in the communications module. Redoing the QA for another module will take too long - Solution: switch to another launch vehicle, will launch from Baikonur (they have the only comms that fit to that size and type of vehicle).
T minus one year: Oops, Baikonur has to trash their main dish, because it needs some crystals that were only manufactured during the Soviet Union, and stores are running out. No problem, manufacture is being revved up again, will only take two years...
Solution (with sweat forming on forehead): Launch two smaller vehicles from Guiana, couple loads in orbit, go to Mars. We did that ten years ago, worked like a charm, didn't it? T minus five months: Oops, that maneuver was on an orbit we cannot currently use due to debris, and the research group that figured out the nitty-gritty ten years ago, and could easily redo their work for another orbit (well, after rewriting most of their code, because that one package turned out to be buggy as hell, and has since been deprecated, and the cloud-platform it ran on has switched gears since) - yeah, those gals all drowned on that boating accident three years ago, remember? No problem, other people are just as smart, and will be easily able to reproduce and test their solution in, say, 6 months?
Solution (with medication-enhanced calmness): Launch with emergency-QA'ed comms on original vehicle.
T minus one week: Ah, chucks, that incoming weather from Cuba is a little too gusty to already put our rocket on the pad, no worries, the windows stays open for like 14 days, we'll just wait it out. One week later: T minus one week: People, do we really only have that one effing crawler to take the rocket out to pad? And you are telling me a vital hydraulics-line broke? What diameter? Are you kidding me? Ok have one hand-knit from unicorn hair, if needs be - there's a one-billion dollar mission waiting on that line - what? two weeks?
...Etc...
Point being that most space mission have one payload, five approximately applicable vehicle-types, one launch facility per type, .... the redundancy is really small because of cost. Everything is super specific, super expensive, super rare, so it pains to keep more than a few of it in stock. There is less than ten chip factories on earth, that could produce certain chips ubiquitous in all the space-equipment, and only two or three are, at any point in time.
For cinematic reasons, i would have the container with the needed goods be stuck on a bottom-rung container on a ship grounded in the Suez Canal...