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

Re: [External] Re: [PATCH 2/2] fetch-pack.c: do not declare local commits as "have" in partial repos

From
韩仰 <hanyang.tony@bytedance.com>
Date
Sep 23, 2024, 03:44 UTC
Message-ID
<CAG1j3zHJVrpK5JZtUXFwkZgWY1-CxqET+ygpaMqo5aM-KeWaxg@mail.gmail.com>
In-Reply-To
<xmqqr09c89id.fsf@gitster.g>
On Sun, Sep 22, 2024 at 2:53 PM Junio C Hamano <gitster@pobox.com> wrote:
Show 7 quoted lines
> I was hoping to see that the issue can be fixed on the "gc" side,
> regardless of how the objects enter our repository, but perhaps I am
> missing something.  Isn't it just the matter of collecting C1, C3
> but not C2?  Or to put it another way, if we first create a list of
> objects to be packed (regardless of whether they are in promisor
> packs), and then remove the objects that are in promisor packs from
> the list, and pack the objects still remaining in the list?

I tried to fix the issue on the "gc" side following JTan's suggestion, by packing local objects referenced by promisor objects into promisor packs. But it turns out the cost for "for each promisor object, parse them and try to decide the objects they reference is in local repo" is too great. In a test blob:none partial clone repo, the gc would take more than one hour in the 2019 MacBook, despite the repo only having 17071073 objects. Normally it would take about 30 minutes.

> if we first create a list of
> objects to be packed (regardless of whether they are in promisor
> packs), and then remove the objects that are in promisor packs from
> the list, and pack the objects still remaining in the list?

This would work, though the remaining objects in the list would be suboptimally packed, due to the delta heuristic. Because we feed object id directly into git-pack-objects, instead of using rev-list. But that's how we pack promisor objects anyway.

In $JOB, we modified git-repack to pack everything into a giant promisor pack if the repo is partially cloned. This basically does the same thing as you suggested, but without the cost of constructing the object list and removing the objects in the promisor packs.

Thanks.
Previous: Junio C HamanoNext: Junio C Hamano
Message 22 of 65 in “revision: fix reachable objects being gc'ed in no blob clone repo”
  1. 0/1 revision: fix reachable objects being gc'ed in no blob clone repoHan Young, Aug 2, 2024
  2. 1/1 revision: don't set parents as uninteresting if exclude promisor objectsHan Young, Aug 2, 2024
  3. Junio C HamanoAug 2, 2024
  4. 韩仰Aug 12, 2024
  5. Junio C HamanoAug 12, 2024
  6. 韩仰Aug 22, 2024
  7. Jonathan TanAug 13, 2024
  8. Jonathan TanAug 13, 2024
  9. Junio C HamanoAug 14, 2024
  10. Jonathan TanAug 14, 2024
  11. 0/4 revision: fix reachable objects being gc'ed in no blob clone repoHan Young, Aug 23, 2024
  12. 1/4 packfile: split promisor objects oidset into twoHan Young, Aug 23, 2024
  13. 2/4 revision: add exclude-promisor-pack-objects optionHan Young, Aug 23, 2024
  14. 3/4 revision: don't mark commit as UNINTERESTING if --exclude-promisor-objects is setHan Young, Aug 23, 2024
  15. 4/4 repack: use new exclude promisor pack objects optionHan Young, Aug 23, 2024
  16. 0/2 revision: fix reachable commits being gc'ed in partial repoCalvin Wan, Sep 19, 2024
  17. 1/2 packfile: split promisor objects oidset into twoCalvin Wan, Sep 19, 2024
  18. Junio C HamanoSep 22, 2024
  19. 2/2 fetch-pack.c: do not declare local commits as "have" in partial reposCalvin Wan, Sep 19, 2024
  20. Junio C HamanoSep 22, 2024
  21. Junio C HamanoSep 22, 2024
  22. 韩仰Sep 23, 2024
  23. Junio C HamanoSep 23, 2024
  24. Calvin WanOct 2, 2024
  25. 0/2 repack: pack everything into promisor packfile in partial reposHan Young, Sep 25, 2024
  26. 1/2 repack: pack everything into packfileHan Young, Sep 25, 2024
  27. 2/2 t0410: adapt tests to repack changesHan Young, Sep 25, 2024
  28. Phillip WoodSep 25, 2024
  29. Junio C HamanoSep 25, 2024
  30. Junio C HamanoSep 25, 2024
  31. Missing Promisor Objects in Partial Repo Design DocCalvin Wan, Oct 1, 2024
  32. Junio C HamanoOct 1, 2024
  33. Junio C HamanoOct 2, 2024
  34. Han YoungOct 2, 2024
  35. Calvin WanOct 8, 2024
  36. Han YoungOct 9, 2024
  37. Jonathan TanOct 9, 2024
  38. Jonathan TanOct 12, 2024
  39. Han YoungOct 12, 2024
  40. Jonathan TanOct 14, 2024
  41. Jonathan TanOct 9, 2024
  42. 0/3 repack: pack everything into promisor packfile in partial reposHan Young, Oct 8, 2024
  43. 1/3 repack: pack everything into packfileHan Young, Oct 8, 2024
  44. Calvin WanOct 8, 2024
  45. 2/3 t0410: adapt tests to repack changesHan Young, Oct 8, 2024
  46. 3/3 partial-clone: update docHan Young, Oct 8, 2024
  47. Junio C HamanoOct 8, 2024
  48. Junio C HamanoOct 8, 2024
  49. Han YoungOct 9, 2024
  50. 0/3 repack: pack everything into promisor packfile in partial reposHan Young, Oct 11, 2024
  51. 1/3 repack: pack everything into packfileHan Young, Oct 11, 2024
  52. 2/3 repack: adapt tests to repack changesHan Young, Oct 11, 2024
  53. 3/3 partial-clone: update docHan Young, Oct 11, 2024
  54. Junio C HamanoOct 11, 2024
  55. Junio C HamanoOct 11, 2024
  56. 0/3 repack: pack everything into promisor packfile in partial reposHan Young, Oct 14, 2024
  57. 1/3 repack: pack everything into packfileHan Young, Oct 14, 2024
  58. 2/3 t0410: adapt tests to repack changesHan Young, Oct 14, 2024
  59. 3/3 partial-clone: update docHan Young, Oct 14, 2024
  60. 0/3 Repack on fetchJonathan Tan, Oct 21, 2024
  61. 1/3 move variableJonathan Tan, Oct 21, 2024
  62. 2/3 pack-objectsJonathan Tan, Oct 21, 2024
  63. 3/3 record local links and call pack-objectsJonathan Tan, Oct 21, 2024
  64. Han YoungOct 23, 2024
  65. Jonathan TanOct 23, 2024

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.