It looks like you need to turn off Use Timestep in Volume Source when working with FLIP.
This is tested on 1, 2, 4 and 8 Dopnet Substeps for collisionvel. With the Use Timestep toggle off, the simulation was 100% correct. With it on, the sliding effect appeared.

When rendering volumetrics, the scattering phase parameter affects how the object will be shaded: with forward or back scattering.
If scattering phase > 0, then forward scattering occurs. This is typical for water fog.
If scattering phase < 0, then back scattering occurs.  This is typical for smoke.

It is advisable to use packed fragments only in cases where either all fragments or most of them are used. A feature of packaged fragments is that “each fragment shares the same geometry, but refers to a subset of it.” Therefore, when deleting individual fragments, there is no memory saving. Even if you delete all the fragments, leaving only one, and write it to disk, then the cache size will be equal to the memory size of the entire model. And when loading the cache, you will still see only one fragment.

In cases where the logic of work involves changing the number of packeds, it is better to use the usual packed geometry. You can create them in a for-each loop with the pack node without breaking them down by name. Also, the copy to points node creates packed geometry.

The behavior of an object during bending with a cone twist constraint is strongly influenced by the physical parameter the rotational stiffness. If it is high, then after an impact the object bends worse.

It is recommended that you always specify attribute data types: i@class, f@pscale, s@name, and so on. There are a large number of possible errors associated with changing the data type of attributes. And initializing the data type is the first step to avoid these errors.

The point radius scale in the VDB From Particles must not be less than:

point radius scale = minimum radius in voxels * voxel size * 1.01

That is, the actual radius of the particle should not be less than the radius of one and a half voxels. In the opposite direction, the voxel size should not be greater than:

voxel size = point radius scale / minimum radius in voxels * 0.99

Using these expressions, you can bind parameters and adaptively set voxel size and point radius.

Try to always use SOP-level nodes in the paths to objects. Otherwise, the render flag will be used for rendering, which generally gives an unpredictable result.
This also applies to a geometric object in a geometry light.

The VDB Smooth SDF produces incorrect values for distance from the surface. If you convert it to geometry and make SDF again, the values will change. Therefore, the result cannot be used for accurate distance calculations (both outside and inside).

When caching geometry to disk, there is no difference between storing a single attribute value on points or on a global attribute.
In the case when, for example, for a large number of points there are only two attribute values, it makes sense to sort by it. This will save storage space. Example: 4.49 GB vs 4.22 GB for a sequence with two values.

If step voxels are visible when rendering the volume, then the necessary density field in the shader is missing.
This often happens when promoting the density attribute from volume shader core. Its name becomes density2 by default, and it must immediately be changed to density.