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

Re: Huge performance bottleneck reading packs

From
Jeff King <peff@peff.net>
Date
Oct 14, 2016, 19:00 UTC
Message-ID
<20161014190051.oxrfreb56nogkdv7@sigill.intra.peff.net>
In-Reply-To
<0e4173b6-8fbb-4223-b061-524c7ebfd4d7@oracle.com>
On Fri, Oct 14, 2016 at 08:55:29AM +0200, Vegard Nossum wrote:
Show 15 quoted lines
> On 10/13/2016 10:43 PM, Jeff King wrote:
> > No problem. I do think you'll benefit a lot from packing everything into
> > a single pack, but it's also clear that Git was doing more work than it
> > needed to be. This was a known issue when we added the racy-check to
> > has_sha1_file(), and knew that we might have to field reports like this
> > one.
> 
> After 'git gc' (with the .5G limit) I have 23 packs and with your patch
> I now get:
> 
> $ time git fetch
> 
> real    0m26.520s
> user    0m3.580s
> sys     0m0.636s

Cool. That's about what I'd expect based on the size numbers you gave earlier. The other 23 seconds are likely being spent on passing the ref advertisement over the network.

-Peff
Previous: Vegard NossumNext: Jakub Narębski
Message 20 of 22 in “Huge performance bottleneck reading packs”
  1. Vegard NossumOct 12, 2016
  2. Junio C HamanoOct 12, 2016
  3. Vegard NossumOct 13, 2016
  4. Jeff KingOct 13, 2016
  5. Jeff KingOct 12, 2016
  6. Jeff KingOct 12, 2016
  7. Jeff KingOct 12, 2016
  8. Vegard NossumOct 13, 2016
  9. Jeff KingOct 13, 2016
  10. fetch: use "quick" has_sha1_file for tag followingJeff King, Oct 13, 2016
  11. Jeff KingOct 13, 2016
  12. Jeff KingOct 13, 2016
  13. Junio C HamanoOct 14, 2016
  14. Jeff KingOct 14, 2016
  15. Junio C HamanoOct 17, 2016
  16. Jeff KingOct 18, 2016
  17. Vegard NossumOct 13, 2016
  18. Jeff KingOct 13, 2016
  19. Vegard NossumOct 14, 2016
  20. Jeff KingOct 14, 2016
  21. Jakub NarębskiOct 14, 2016
  22. Vegard NossumOct 13, 2016

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.