Typical error messages include:
Warning message:
zero-length arrow is of indeterminate angle and so skipped
null device
1

These come from the statistics package R, not from MLOGD. They may be safely ignored.
Device png256 is not available

This comes from the statistics package R, not from MLOGD. It means that your version of R doesn't recognize png256 and, as a result, fails to produce the png versions of the plots. However, you should still get the eps and pdf versions of the plots.
Aborting: sequence seq.003.fasta, unknown nucleotide 'z', at 2345.

The programmes only accept ACGT and standard ambiguous nucleotide codes MRWVHDBSYKNX, upper or lower case, plus '-' and '.' for gaps. seq.003.fasta is the 3rd sequence in the input alignment.
Fitting number of mutations (87) outside that allowed by given t range. Sequences 2 and 4.
Iteration on t failed to converge in 20 iterations. Sequences 4 and 5.

These error messages refer to fitting the t (pairwise sequence divergence) parameter. Generally the results should be fine, even if you get error messages. However you could check whether you have a bad alignment - perhaps one of your sequences is too divergent from the others.
Warning: gap within null model coding region not a multiple of three nt; sequence 3, sequence coords 1280, alignment coords 1296.
Warning: gap within alternative model coding region not a multiple of three nt; sequence 3, sequence coords 1280, alignment coords 1296.

There may be local problems with the various statistics and stop/start codon annotation, as codons may be picked from the wrong read-frame. Provided there aren't too many of these errors, the results should still be basically OK. Otherwise you may need to redo your alignment, insuring that, as far as possible, gaps within coding sequences occur in groups of three. You can use the alignment site to do this.
Warning: orfs.1 has overlapping ORFs. Only the first will be used in overlap regions.

If either the input 'Known CDS(s)' (null model, orfs.1) or input 'Query CDS(s)' (alternate model, orfs.2) contains overlapping ORFs then, in the overlap regions, only the first ORF will be used in the respective model.
Warning: sequence seq.001.fasta, ambiguous nt 'm', at 5.

The ambiguous nucleotide, and any codons containing the ambiguous nucleotide, will be omitted from the calculation of statistics, but there should be no other ill effects.
Warning: failed to reach requested number of mutations; sequence: 99.

This warning may occur in the Monte Carlo simulation programme for very divergent sequences. Mutations are applied until the required number of mutations are accepted, but there is a time-out switch in case obtaining the required number of accepted mutations takes too long.
More serious error messages include:
Aborting: ORF 3 in orfs.1, endpoints outside sequence range.

Check the CDS annotation. Remember that these should be in the ungapped reference sequence coordinates, not alignment coordinates. orfs.1 refers to the 'Known CDS(s)' input while orfs.2 refers to the 'Query CDS(s)' input.
Warning: number of nucleotides in ORF 1 in orfs.2 not a multiple of 3.

The programme will continue to work, but you should check your CDS annotation. Note that for frameshifts, disjoint CDSs and circular genomes, you should use the NCBI GenBank 'join(2307..3212,1..1623)' format in the CDS annotation.
Aborting: gapped sequences of differing length.

Check your input alignment. All sequences, when gap characters are included, must be of the same length.
Detected an ambiguous nt code in the reference sequence. Can't run Monte Carlo simulations in this situation.

The Monte Carlo simulation can not run if the initial sequence (the reference sequence) contains ambiguous nucleotide codes. Choose a different reference sequence, or else fudge the results by picking a random nucleotide (A, C, G or T) for each ambiguous nucleotide in the reference sequence.