git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH 4/6] t/perf: add perf tests for fetches from a bitmapped server

From
Jeff King <peff@peff.net>
Date
Aug 17, 2018, 20:57 UTC
Message-ID
<20180817205741.GD20088@sigill.intra.peff.net>
In-Reply-To
<20180817205427.GA19580@sigill.intra.peff.net>

A server with bitmapped packs can serve a clone very quickly. However, fetches are not necessarily made any faster, because we spend a lot less time in object traversal (which is what bitmaps help with) and more time finding deltas (because we may have to throw out on-disk deltas if the client does not have the base).

As a first step to making this faster, this patch introduces a new perf script to measure fetches into a repo of various ages from a fully-bitmapped server.

We separately measure the work done by the server (in pack-objects) and that done by the client (in index-pack). Furthermore, we measure the size of the resulting pack.

Breaking it down like this (instead of just doing a regular "git fetch") lets us see how much each side benefits from any changes. And since we know the pack size, if we estimate the network speed, then one could calculate a complete wall-clock time for the operation (though the script does not do this automatically).

Signed-off-by: Jeff King <peff@peff.net>
---
 t/perf/p5311-pack-bitmaps-fetch.sh | 45 ++++++++++++++++++++++++++++++
 1 file changed, 45 insertions(+)
 create mode 100755 t/perf/p5311-pack-bitmaps-fetch.sh
diff --git a/t/perf/p5311-pack-bitmaps-fetch.sh b/t/perf/p5311-pack-bitmaps-fetch.sh
new file mode 100755
index 0000000000..b04575951f
--- /dev/null
+++ b/t/perf/p5311-pack-bitmaps-fetch.sh
@@ -0,0 +1,45 @@
+#!/bin/sh
+
+test_description='performance of fetches from bitmapped packs'
+. ./perf-lib.sh
+
+test_perf_default_repo
+
+test_expect_success 'create bitmapped server repo' '
+	git config pack.writebitmaps true &&
+	git config pack.writebitmaphashcache true &&
+	git repack -ad
+'
+
+# simulate a fetch from a repository that last fetched N days ago, for
+# various values of N. We do so by following the first-parent chain,
+# and assume the first entry in the chain that is N days older than the current
+# HEAD is where the HEAD would have been then.
+for days in 1 2 4 8 16 32 64 128; do
+	title=$(printf '%10s' "($days days)")
+	test_expect_success "setup revs from $days days ago" '
+		now=$(git log -1 --format=%ct HEAD) &&
+		then=$(($now - ($days * 86400))) &&
+		tip=$(git rev-list -1 --first-parent --until=$then HEAD) &&
+		{
+			echo HEAD &&
+			echo ^$tip
+		} >revs
+	'
+
+	test_perf "server $title" '
+		git pack-objects --stdout --revs \
+				 --thin --delta-base-offset \
+				 <revs >tmp.pack
+	'
+
+	test_size "size   $title" '
+		wc -c <tmp.pack
+	'
+
+	test_perf "client $title" '
+		git index-pack --stdin --fix-thin <tmp.pack
+	'
+done
+
+test_done
-- 
2.18.0.1205.g3878b1e64a
Previous: Jeff KingNext: Jeff King
Message 5 of 17 in “reuse on-disk deltas for fetches with bitmaps”
  1. 0/6 reuse on-disk deltas for fetches with bitmapsJeff King, Aug 17, 2018
  2. 1/6 t/perf: factor boilerplate out of test_perfJeff King, Aug 17, 2018
  3. 2/6 t/perf: factor out percent calculationsJeff King, Aug 17, 2018
  4. 3/6 t/perf: add infrastructure for measuring sizesJeff King, Aug 17, 2018
  5. 4/6 t/perf: add perf tests for fetches from a bitmapped serverJeff King, Aug 17, 2018
  6. 5/6 pack-bitmap: save "have" bitmap from walkJeff King, Aug 17, 2018
  7. Stefan BellerAug 17, 2018
  8. Jeff KingAug 17, 2018
  9. 6/6 pack-objects: reuse on-disk deltas for thin "have" objectsJeff King, Aug 17, 2018
  10. Stefan BellerAug 17, 2018
  11. Jeff KingAug 17, 2018
  12. Junio C HamanoAug 20, 2018
  13. Jeff KingAug 20, 2018
  14. 0/6 reuse on-disk deltas for fetches with bitmapsJeff King, Aug 21, 2018
  15. Jeff KingAug 21, 2018
  16. Junio C HamanoAug 21, 2018
  17. Jeff KingAug 21, 2018

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.