Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 448462 - speedup repoman by using parallel jobs
Summary: speedup repoman by using parallel jobs
Status: RESOLVED WONTFIX
Alias: None
Product: Portage Development
Classification: Unclassified
Component: Repoman (show other bugs)
Hardware: All Linux
: Normal enhancement
Assignee: Portage team
URL:
Whiteboard:
Keywords: PATCH
Depends on: 736473 736334
Blocks:
  Show dependency tree
 
Reported: 2012-12-24 22:02 UTC by Sergei Trofimovich (RETIRED)
Modified: 2022-07-12 03:18 UTC (History)
1 user (show)

See Also:
Package list:
Runtime testing required: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Sergei Trofimovich (RETIRED) gentoo-dev 2012-12-24 22:02:08 UTC
repoman is mostly a cpu-bound task even with generated metadata.


[sf] ~/portage/gentoo-haskell:time ~/portage/gentoo-qa/run_repoman.sh 

    real    5m20.469s
    user    4m35.703s
    sys     0m43.763s

[sf] ~/portage/gentoo-haskell:cat ~/portage/gentoo-qa/run_repoman.sh
    #!/bin/sh

    repoman full -d > repoman-QA-`date +%F-%T`.log

Would be nice to make repoman use of --jobs/EJOBS variables to speed
things down a multicore boxes.

Thanks!
Comment 1 Zac Medico gentoo-dev 2012-12-25 20:49:27 UTC
Given the current design, doing one fork per package ($CATEGORY/$PN) would make sense, since it would allow dependency match caches to be shared between multiple versions of the same package. We'd just have to split out the dependency checking part (which consumes the bulk of the cpu time) and run it from within instances of portage's ForkProcess class. The results could be sent back to the main process via a pipe, similarly to how our MergeProcess class sends back elog messages.
Comment 2 Jonas Stein gentoo-dev 2017-11-14 11:10:26 UTC
as workaround one could run repoman with 
sys-process/parallel
until this is fixed.
Comment 4 Larry the Git Cow gentoo-dev 2020-08-17 05:32:26 UTC
The bug has been referenced in the following commit(s):

https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=89ceaeb6b4619dcddace5dc70a126c360ccdd8fe

commit 89ceaeb6b4619dcddace5dc70a126c360ccdd8fe
Author:     Zac Medico <zmedico@gentoo.org>
AuthorDate: 2020-08-17 05:18:21 +0000
Commit:     Zac Medico <zmedico@gentoo.org>
CommitDate: 2020-08-17 05:32:14 +0000

    sys-apps/portage: Bump to version 3.0.3
    
     #448462 repoman: avoid "RuntimeError: Event loop is closed"
             after SIGINT with --jobs > 1.
     #699256 emerge: Remove deprecated --changelog option
     #699256 emerge: Add short -l option for --load-average
     #711174 SpawnProcess: Fix _cancel to cleanup PipeLogger quickly
    
    Bug: https://bugs.gentoo.org/733180
    Bug: https://bugs.gentoo.org/448462
    Bug: https://bugs.gentoo.org/699256
    Bug: https://bugs.gentoo.org/711174
    Package-Manager: Portage-3.0.3, Repoman-2.3.23
    Signed-off-by: Zac Medico <zmedico@gentoo.org>

 sys-apps/portage/Manifest             |   1 +
 sys-apps/portage/portage-3.0.3.ebuild | 263 ++++++++++++++++++++++++++++++++++
 2 files changed, 264 insertions(+)
Comment 5 Larry the Git Cow gentoo-dev 2020-08-17 06:13:17 UTC
The bug has been referenced in the following commit(s):

https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=09578270a9cdb7ad6ed533b4f3dec816c5ab8868

commit 09578270a9cdb7ad6ed533b4f3dec816c5ab8868
Author:     Zac Medico <zmedico@gentoo.org>
AuthorDate: 2020-08-17 05:59:17 +0000
Commit:     Zac Medico <zmedico@gentoo.org>
CommitDate: 2020-08-17 06:12:54 +0000

    app-portage/repoman: Bump to version 3.0.0
    
     #448462 Add --jobs and --load-average options which allow dependency
             checks for multiple profiles to run in parallel
    
    Bug: https://bugs.gentoo.org/448462
    Package-Manager: Portage-3.0.3, Repoman-3.0.0
    Signed-off-by: Zac Medico <zmedico@gentoo.org>

 app-portage/repoman/Manifest             |  1 +
 app-portage/repoman/repoman-3.0.0.ebuild | 64 ++++++++++++++++++++++++++++++++
 2 files changed, 65 insertions(+)
Comment 6 Larry the Git Cow gentoo-dev 2020-08-17 06:13:31 UTC
The bug has been referenced in the following commit(s):

https://gitweb.gentoo.org/proj/portage.git/commit/?id=26524e5fbc8019e22d15765a35570df7c7a315ca

commit 26524e5fbc8019e22d15765a35570df7c7a315ca
Author:     Zac Medico <zmedico@gentoo.org>
AuthorDate: 2020-08-15 02:51:38 +0000
Commit:     Zac Medico <zmedico@gentoo.org>
CommitDate: 2020-08-17 05:39:06 +0000

    repoman: Add --jobs and --load-average options (bug 448462)
    
    Add --jobs and --load-average options which allow dependency checks
    for multiple profiles to run in parallel. The increase in performance
    is non-linear for the number of jobs, but it can be worthwhile
    (I measured a 35% decrease in time when running 'repoman -j8 full'
    on sys-apps/portage). For the -j1 case (default), all dependency
    checks run in the main process as usual, so there is no significant
    performance penalty for the default case.
    
    Bug: https://bugs.gentoo.org/448462
    Signed-off-by: Zac Medico <zmedico@gentoo.org>

 repoman/lib/repoman/argparser.py                   |   9 ++
 repoman/lib/repoman/modules/scan/depend/profile.py | 117 +++++++++++++++++----
 repoman/man/repoman.1                              |   9 +-
 3 files changed, 116 insertions(+), 19 deletions(-)
Comment 7 Zac Medico gentoo-dev 2020-08-17 07:07:36 UTC
I've got support to run the dependency checks for multiple ebuilds in parallel, but  it needs some work because it's not showing a performance improvement:

https://github.com/gentoo/portage/compare/master...zmedico:bug_448462_parallel_repoman
Comment 8 Sam James archtester Gentoo Infrastructure gentoo-dev Security 2022-07-12 03:18:31 UTC
repoman support has been removed per bug 835013.

Please file a new bug (or, I suppose, reopen this one) if you feel this check is still applicable to pkgcheck and doesn't already exist.