Skip to main content

Public

Project privacy set to public. By default, its content is available to everyone (authenticated or not). Please note that more restrictive permissions might exist on some items.

    •  
      request #3242 Appearance of superfluous push operations when migrating a repo to Gerrit
    Actions
    Infos
    #3242
    Patrick Renaud (patrick.renaud)
    2013-05-28 16:51
    2013-05-01 20:28
    2063
    Details
    Appearance of superfluous push operations when migrating a repo to Gerrit
    The trace of gerrit ssh operations initiated by Tuleap when migrating a repo ends with three successive push operations:

    [2013-05-01 11:45:56,873 -0400] 0e715341 tuleap-gerrit-admin-user a/1000001 LOGIN FROM my-tuleap-test-server
    [2013-05-01 11:45:57,759 -0400] 0e715341 tuleap-gerrit-admin-user a/1000001 'git-receive-pack patrick/repo3' 3ms 524ms 0
    [2013-05-01 11:45:58,030 -0400] 0e715341 tuleap-gerrit-admin-user a/1000001 LOGOUT
    [2013-05-01 11:45:58,984 -0400] 6ea9efb2 tuleap-gerrit-admin-user a/1000001 LOGIN FROM my-tuleap-test-server
    [2013-05-01 11:45:59,731 -0400] 6ea9efb2 tuleap-gerrit-admin-user a/1000001 'git-receive-pack patrick/repo3' 4ms 387ms 0
    [2013-05-01 11:46:00,008 -0400] 6ea9efb2 tuleap-gerrit-admin-user a/1000001 LOGOUT
    [2013-05-01 11:46:00,958 -0400] ee94ff6f tuleap-gerrit-admin-user a/1000001 LOGIN FROM my-tuleap-test-server
    [2013-05-01 11:46:01,456 -0400] ee94ff6f tuleap-gerrit-admin-user a/1000001 'git-receive-pack patrick/repo3' 3ms 134ms 0
    [2013-05-01 11:46:01,696 -0400] ee94ff6f tuleap-gerrit-admin-user a/1000001 LOGOUT

    A single push operation from Tuleap to Gerrit should probably be sufficient to send all the existing data to Gerrit. Although the above seems a bit excessive it causes no apparent harm.

    Can you descrive the reason behind this?
    SCM/Gerrit
    5.12
    Empty
    • [ ] enhancement
    • [ ] internal improvement
    patricia.carrasco@ericsson.com, vladimir.cantiru@ericsson.com, Emilio Palmiero (empa)
    Stage
    Manuel Vacelet (vaceletm)
    Closed
    2013-05-28
    Attachments
    Artifact links
    Empty
    Empty
    References
    References list is empty
    Display settings

    Follow-ups

    User avatar
    Manuel Vacelet (vaceletm)2013-05-28 16:50
    • Status changed from Waiting for information to Verified
    • Assigned to changed from None to Manuel Vacelet (vaceletm)
    User avatar
    Patrick Renaud (patrick.renaud)2013-05-28 16:14
    Yes it does. It confirms my suspicions. All is good then! :-)

    Thank you Manuel. You may close this ticket now.
    User avatar
    Manuel Vacelet (vaceletm)2013-05-28 12:16
    Hi Patrick,

    The first push is for refs/meta/config
    The second push is for branches (refs/heads)
    The third push is for tags (refs/tags)

    Does it answer the question ?
    User avatar
    Patrick Renaud (patrick.renaud)2013-05-21 18:21
    Hi,

    Each command below results in a single 'git-receive-pack' operation on the server; this accounts for two such operations. We need to understand the source of the third (or first!) incantation of it.

    I suspect it may occur as a result of pushing a change to the ACL (refs/meta/config). Can you please confirm if my assumption is correct?
    • Is an Enhancement or an internal improvement? set to
    User avatar
    dylan bowden (dylan)2013-05-16 15:47
    Hi,

    From online documentation such as http://manpages.ubuntu.com/manpages/dapper/man1/git-receive-pack.1.html, I understand that 'git-receive-pack' is a command that is not called directly by a user but rather by 'git push' indirectly. Having looked at the Tuleap sources, there are indeed no instances of 'git-receive-pack' in the code and specifically not within the repository migration part.
    Thus, I began looking for instances of 'git push'. There are indeed multiple git pushes for various refs within the Tuleap migration code:

    git push $gerrit_project_url refs/heads/*:refs/heads/*; git push $gerrit_project_url refs/tags/*:refs/tags/*

    It is plausible that, through the inner-workings of git, one of these two actions results in a double instance of 'git-receive-pack'.

    Do you wish for us to dig deeper or is that enough information?
    Dylan