The problem is possibly two failure modes going on simultaneously.
1. Stainless Steels (18-8) are very prone to fatigue failure --- engine generated vibration that exceeds 1 million load cycles at greater than 30% ultimate tensile strength of the component - the fatigue endurance limit of common stainless steels.
Such exhaust "dry stacks" are usually solely attached to the engine through a pipe adapter flange at the outlet of the exhaust manifold; the dry stack 'sticks up' from the engine similar to a 'flag pole'. This causes the structure to behave as a 'cantilever' whose SHAPE in complex beam structure formulary calculations reduces the inherent strength by approximately 1/4th at the 'root' of the cantilever --- where the dry stack connects to the engine exhaust manifold --- it's the typical failure point for large fatigue cracks. This failure is typical of cantilevers in dynamic situations.
The Rx for this is to support that dry stack at near its top with 'triangular bracing' to the engine ... could be just simple bent pipe, flattened and bent at its ends, attached to near midway to the top of the dry stack with simple hose/muffler clamps and attached onto the one of the cylinder head bots or any available large bolt on the aft vertical face of the 'head' (but NOT on the transmission). The 'triangular' shape of the drystack/engine/bracing is important, the wider the base of that 'triangle, the better. This arrangement will tend to make the top of the dry stack vibrate at the same frequency as the 'root' connection to the engine for less 'flex' at the stack to engine connection. Without the bracing, that 'contraption' sticking up and vibrating independently is a force multiplying LEVER.
2. Stainless is prone to stress corrosion ... in this case, from the sulphur oxides produced by the combustion of diesel fuel. Such combustion generates water and sulphur oxides which then form various sulphur and nitrogen derived acids: sulphur dioxide + water, HOT sulphurous and sulphuric acids, nitrous/nitric acids etc. If there are micro-cracks in the surface caused by the generation of fatigue failure these potent acids then penetrate progressively deeper into the metal via the micro-cracks causing deeper and deeper erosion and which in turn produce 'faster' fatigue, etc. which produces deeper penetration via chemical attack, etc. etc. etc.
A two phase failure!
1. stop the top of the dry stack from vibrating out of phase with the engine by 'tying' it to the engine with a 'robust' triangular support. 3/4" to 1" thin walled stainless pipe 'will do'.
2. replace the stainless steel components (as they fail) with simple threaded moderately thick walled black iron piping. Black iron is the standard for such exhaust applications:
A. better fatigue resistance; and better high temperature service life due to the formation of 'protective' iron oxides on the internal surface due to the 'heat'.
B. better chemical resistance to exhaust gases due to less propensity to form surface micro-cracks due to better 'fatigue limit' of the material.
C. quite 'cheap' as is made from 'plumbing supply' parts, as well as easily replaceable.
Note - threaded cast iron exhaust manifold to 'dry stack' adapters are readily available at most Yanmar dealers, etc.
I realize that this response is 'quite technical''; so, if you need a simpler explanation, etc. just PM me.