mononoke: make chunk size configurable in regenerate_filenodes binary

Summary:
It's useful to have it configurable.
While here, also use slog instead of println to attach timestamp as well

Reviewed By: Croohand

Differential Revision: D29165693

fbshipit-source-id: d844926560b15042445d5861a281870ac102d12e

eden/mononoke/regenerate_hg_filenodes/Cargo.toml

@@ -18,6 +18,7 @@ derived_data_filenodes = { version = "0.1.0", path = "../derived_data/filenodes"
 fbinit = { version = "0.1.0", git = "https://github.com/facebookexperimental/rust-shed.git", branch = "master" }
 futures = { version = "0.3.13", features = ["async-await", "compat"] }
 mercurial_types = { version = "0.1.0", path = "../mercurial/types" }
+slog = { version = "2.5", features = ["max_level_trace"] }
 
 [patch.crates-io]
 addr2line = { git = "https://github.com/gimli-rs/addr2line.git", rev = "0b6b6018b5b252a18e628fba03885f7d21844b3c" }

eden/mononoke/regenerate_hg_filenodes/src/main.rs

@@ -21,6 +21,7 @@ use fbinit::FacebookInit;
 use futures::compat::Future01CompatExt;
 use futures::future::{join_all, FutureExt};
 use mercurial_types::{HgChangesetId, HgNodeHash};
+use slog::info;
 use std::fs::File;
 use std::str::FromStr;
 use std::{
@@ -66,9 +67,9 @@ async fn regenerate_all_manifests(
     ctx: CoreContext,
     repo: BlobRepo,
     hg_css: Vec<HgChangesetId>,
+    chunk_size: usize,
 ) -> Result<(), Error> {
     let mut i = 0;
-    let chunk_size = 100;
     for chunk in hg_css.chunks(chunk_size) {
         let mut futs = vec![];
         for hg_cs in chunk {
@@ -77,11 +78,11 @@ async fn regenerate_all_manifests(
                 repo.clone(),
                 *hg_cs,
             ));
+            i += 1;
         }
         let res: Result<Vec<_>, Error> = join_all(futs).await.into_iter().collect();
         res?;
-        i += chunk_size;
-        println!("processed {}", i);
+        info!(ctx.logger(), "processed {}", i);
     }
     Ok(())
 }
@@ -94,6 +95,7 @@ fn main(fb: FacebookInit) -> Result<(), Error> {
             r#"
                --i-know-what-i-am-doing
                --file [FILE]                        'contains list of commit hashes for which we want to regenerate filenodes'
+               --chunk-size [SIZE]                  'how many commits to process at once (default: 100)'
             "#,
         )
         .get_matches(fb)?;
@@ -112,6 +114,8 @@ fn main(fb: FacebookInit) -> Result<(), Error> {
 
     let inputfile = matches.value_of("file").expect("input file is not set");
     let inputfile = File::open(inputfile).expect("cannot open input file");
+
+    let chunk_size = args::get_usize(&matches, "chunk-size", 100);
     let file = BufReader::new(&inputfile);
     rt.block_on(
         regenerate_all_manifests(
@@ -120,6 +124,7 @@ fn main(fb: FacebookInit) -> Result<(), Error> {
             file.lines()
                 .map(|line| convert_to_cs(&line.unwrap()).unwrap())
                 .collect(),
+            chunk_size,
         )
         .boxed(),
     )

eden/mononoke/tests/integration/test-gitimport-octopus.t

@@ -113,7 +113,7 @@
   $ hg log -r 'all()' -T '{node}\n' > hashes
   $ regenerate_hg_filenodes --file 'hashes'
   * using repo "repo" repoid RepositoryId(0) (glob)
-  processed 100
+  * processed 5 (glob)
 
   $ sqlite3 "$TESTTMP/monsql/sqlite_dbs" "SELECT HEX(filenode), HEX(linknode) FROM filenodes ORDER BY filenode DESC;"
   DDAE7A95B6B0FB27DFACC4051C41AA9CFF30C1E2|389710B7D0BE8C40323D3D96F717219F1CCE9601