Cleanup old folders on central

regression_tests/move_output.jl

@@ -9,11 +9,11 @@ job_ids = getindex.(split.(lines, "\""), 2)
 @assert count(x -> occursin("OrderedDict", x), all_lines) == length(job_ids) + 1
 @assert length(job_ids) ≠ 0 # safety net
 
+# Note: cluster_data_prefix is also defined in compute_mse.jl
+cluster_data_prefix = "/central/scratch/esm/slurm-buildkite/climaatmos-main"
 if get(ENV, "BUILDKITE_PIPELINE_SLUG", nothing) == "climaatmos-ci"
     commit = ENV["BUILDKITE_COMMIT"]
     branch = ENV["BUILDKITE_BRANCH"]
-    # Note: cluster_data_prefix is also defined in compute_mse.jl
-    cluster_data_prefix = "/central/scratch/esm/slurm-buildkite/climaatmos-main"
 
     @info "pwd() = $(pwd())"
     @info "branch = $(branch)"
@@ -46,3 +46,47 @@ if get(ENV, "BUILDKITE_PIPELINE_SLUG", nothing) == "climaatmos-ci"
 else
     @info "ENV keys: $(keys(ENV))"
 end
+
+function get_ref_counter(p)
+    f = joinpath(p, "ref_counter.jl")
+    if !isfile(f)
+        @warn "Expected ref counter file did not exist in path $p\n path contents:\n$(readdir(p))"
+        return nothing
+    end
+    return parse(Int, first(readlines(f)))
+end
+function cleanup_central(cluster_data_prefix)
+    @warn "Cleaning up old files on central"
+    # Get (sorted) array of paths, `pop!(sorted_paths)`
+    # is the most recent merged folder.
+    sorted_paths = sorted_dataset_folder(; dir = cluster_data_prefix)
+    keep_latest_n = 10
+    keep_latest_ref_counters = 5
+    if !isempty(sorted_paths)
+        N = length(sorted_paths) - keep_latest_n
+        paths_to_delete = []
+        for i in 1:N
+            ref_counter = get_ref_counter(sorted_paths[i])
+            ref_counter_last = get_ref_counter(sorted_paths[end])
+            if isnothing(ref_counter_last) || isnothing(ref_counter)
+                @show sorted_paths
+                error("Reference counters did not exist")
+            end
+            # Just to be safe, let's also make sure that we don't delete
+            # any paths with recent (let's say 5) ref counter increments ago.
+            if ref_counter + keep_latest_ref_counters < ref_counter_last
+                push!(paths_to_delete, sorted_paths[i])
+            end
+        end
+        @info "Deleting files:"
+        for i in 1:N
+            f = paths_to_delete[i]
+            @info "     (File, date): ($(f), $(Dates.unix2datetime(stat(f).mtime)))"
+        end
+        # for i in 1:N
+        #     rm(paths_to_delete[i])
+        # end
+    end
+end
+
+cleanup_central(cluster_data_prefix)