| Message ID | cover.1733170252.git.jonathantanmy@google.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Performance improvements for repacking non-promisor objects | expand |
On 2024.12.02 12:18, Jonathan Tan wrote: > This is a follow-up to jt/repack-local-promisor [1] (but these patches > are based on master, since that branch has already been merged). > > These patches speed up a fetch that takes 7 hours to take under 3 > minutes. More details are in the commit messages, especially that of > patch 1. > > Thanks in advance to everyone who reviews. While review is going on, > we'll also be testing these at $DAYJOB (I've tested it to work on one > known big repo, but there may be others). > > [1] https://lore.kernel.org/git/cover.1730491845.git.jonathantanmy@google.com/ > > Jonathan Tan (3): > index-pack: dedup first during outgoing link check > index-pack: no blobs during outgoing link check > index-pack: commit tree during outgoing link check > > builtin/index-pack.c | 49 +++++++++++++++++++++++--------------------- > 1 file changed, 26 insertions(+), 23 deletions(-) > > -- > 2.47.0.338.g60cca15819-goog > > This series looks good to me, thanks! Reviewed-by: Josh Steadmon <steadmon@google.com>
Josh Steadmon <steadmon@google.com> writes: >> Thanks in advance to everyone who reviews. While review is going on, >> we'll also be testing these at $DAYJOB (I've tested it to work on one >> known big repo, but there may be others). >> >> [1] https://lore.kernel.org/git/cover.1730491845.git.jonathantanmy@google.com/ >> >> Jonathan Tan (3): >> index-pack: dedup first during outgoing link check >> index-pack: no blobs during outgoing link check >> index-pack: commit tree during outgoing link check >> >> builtin/index-pack.c | 49 +++++++++++++++++++++++--------------------- >> 1 file changed, 26 insertions(+), 23 deletions(-) >> >> -- >> 2.47.0.338.g60cca15819-goog >> >> > > This series looks good to me, thanks! > > Reviewed-by: Josh Steadmon <steadmon@google.com> Thanks.
Jonathan Tan <jonathantanmy@google.com> writes: > This is a follow-up to jt/repack-local-promisor [1] (but these patches > are based on master, since that branch has already been merged). > > These patches speed up a fetch that takes 7 hours to take under 3 > minutes. More details are in the commit messages, especially that of > patch 1. > > Thanks in advance to everyone who reviews. While review is going on, > we'll also be testing these at $DAYJOB (I've tested it to work on one > known big repo, but there may be others). > > [1] https://lore.kernel.org/git/cover.1730491845.git.jonathantanmy@google.com/ > > Jonathan Tan (3): > index-pack: dedup first during outgoing link check > index-pack: no blobs during outgoing link check > index-pack: commit tree during outgoing link check > > builtin/index-pack.c | 49 +++++++++++++++++++++++--------------------- > 1 file changed, 26 insertions(+), 23 deletions(-) When merged to 'seen', this seems to break quite a many tests, all related to "pack". I haven't tried running tests on the topic stand-alone. Test Summary Report ------------------- t5310-pack-bitmaps.sh (Wstat: 256 (exited 1) Tests: 230 Failed: 3) Failed tests: 26, 100, 179 Non-zero exit status: 1 t5327-multi-pack-bitmaps-rev.sh (Wstat: 256 (exited 1) Tests: 314 Failed: 6) Failed tests: 25, 70, 136, 182, 227, 293 Non-zero exit status: 1 t5616-partial-clone.sh (Wstat: 256 (exited 1) Tests: 47 Failed: 3) Failed tests: 5, 38-39 Non-zero exit status: 1 t5326-multi-pack-bitmaps.sh (Wstat: 256 (exited 1) Tests: 357 Failed: 6) Failed tests: 25, 70, 146, 199, 244, 320 Non-zero exit status: 1 t0410-partial-clone.sh (Wstat: 256 (exited 1) Tests: 38 Failed: 4) Failed tests: 11, 14-15, 38 Non-zero exit status: 1 t6020-bundle-misc.sh (Wstat: 256 (exited 1) Tests: 30 Failed: 4) Failed tests: 16-19 Non-zero exit status: 1 t9211-scalar-clone.sh (Wstat: 256 (exited 1) Tests: 13 Failed: 2) Failed tests: 4-5 Non-zero exit status: 1 Files=1031, Tests=31721, 324 wallclock secs (11.58 usr 4.57 sys + 674.15 cusr 6136.28 csys = 6826.58 CPU) Result: FAIL g
Junio C Hamano <gitster@pobox.com> writes: > When merged to 'seen', this seems to break quite a many tests, all > related to "pack". > > I haven't tried running tests on the topic stand-alone. Ah, sorry for a false alarm. The breakage may be coming from some other topic. Haven't figured out which one yet.
Junio C Hamano <gitster@pobox.com> writes: > Junio C Hamano <gitster@pobox.com> writes: > >> When merged to 'seen', this seems to break quite a many tests, all >> related to "pack". >> >> I haven't tried running tests on the topic stand-alone. > > Ah, sorry for a false alarm. The breakage may be coming from some > other topic. Haven't figured out which one yet. Sorry for a double false alarm. The breakages are due to this topic. These do not break without this topic in 'seen', and do break with this topic in 'seen'.