On Tue, Sep 12, 2017 at 7:14 PM, Michael Han <[email protected]> wrote:
> >> "Here the top of commit has commit has c14cdd3673ebdd8014118ec961068e > e51f2327b1, > instead of a1b9460dbfaf9e96aa958d805cf6cf11c8d01d0f, which is expected. " > is > one of the things which lead me to my conclusions. > > This was referring to two different commits, instead of the same commit > that mutated to two different commit hashes. > > >> It's one of the reasons why I thought/think force pushes should be > disallowed > > This was done in https://issues.apache.org/jira/browse/INFRA-13968, no > worries moving forward. > > >> I'm not sure we'll ever know at this point. > > I think it would be good to know what caused this. My bet is on the svn-git > move still, so I left a comment on > https://issues.apache.org/jira/browse/INFRA-12573, let's see what Infra > says about this. > Though, we probably can't do anything about this (like revert to old commit > hash), and the current status might be OK as it's not malicious. > > I did some further research on this and it seems like the SVN->GIT is definitely a possibility for when this happened (I don't see any updates on the INFRA jira, Michael's comment, fwiw). The move was on Apr17th. I checked some of the commits via the records on [email protected] and they are consistent after 4/17/17. I also checked both mine and Lars's repos (which don't correlate in terms of hashes as I stated earlier) and in both cases those repos have commits only prior to 4/17, meaning they are cloned from the "pre svn migration" repos. I suspect that as part of the SVN migration there was the glitch I noted earlier (author name missing) which skewed the Hashes. afaict (see my comments earlier about checking the contents of the repos) this is not due to any malfeasance. At least as far as I can tell. Patrick > > > On Tue, Sep 12, 2017 at 1:12 PM, Patrick Hunt <[email protected]> wrote: > > > On Tue, Sep 12, 2017 at 12:19 PM, Michael Han <[email protected]> wrote: > > > > > Hi Pat, > > > > > > I still think it is the svn git migration - or at least something other > > > than, or in combination with the force push commit that causes this. > One > > > proof is, one of the forked repo already has commit hash difference > > > comparing to the old repos: > > > > > > Forked Repo: > > > https://github.com/rakeshadr/zookeeper-1/commit/ > > > d497aac4cf8cf2144a377e46011385b20fc74fa6 > > > Old Repo: > > > https://github.com/autodesk-forks/zookeeper/commit/ > > > ebd64c4a9b62876f7054add6a07bd36581916557 > > > > > > And this commit was made before we did the svn-git move. > > > > > > Please also note that the forked repo here does not contain the forced > > push > > > commit - so the commit has difference was created before the forced > push > > > happened. For reference, the forced push commit was to reset the HEAD > (at > > > the time that commit was made) to this one: > > > https://github.com/apache/zookeeper/commit/ > > c14cdd3673ebdd8014118ec961068e > > > e51f2327b1 > > > > > > If on the other side it's the forced commit push that causes this, we > > > should expect the same commit hashes for the commits I mentioned above > > > right? > > > > > > > > Based on what I've seen so far I don't know what happened. However your > > comment on the jira here > > https://issues.apache.org/jira/browse/INFRA-13916 > > "Here the top of commit has commit has > > c14cdd3673ebdd8014118ec961068ee51f2327b1, instead of > > a1b9460dbfaf9e96aa958d805cf6cf11c8d01d0f, which is expected. " > > is one of the things which lead me to my conclusions. > > > > I'm not sure we'll ever know at this point. It's one of the reasons why I > > thought/think force pushes should be disallowed, in particular because > it's > > so hard to know what happened when and track the impact. > > > > From what I can see at this point in time, while the hashes differ it > looks > > like it's not malicious, which is my primary concern. I diffed some of > the > > branches and while the hash differs I have not see diffs in the code > > itself. My testing has not be exhaustive however. > > > > Patrick > > > > > > > On Tue, Sep 12, 2017 at 11:15 AM, Patrick Hunt <[email protected]> > wrote: > > > > > > > I suspect it was during the force push. All the commits up to > > > > 11d2d6fd92acf9abc762c41e0f7b91c5acc89f4f -- in all the repos I > listed > > > > including the "old/original" ones -- have the same hash. Thereafter > > they > > > > differ. Look at the following comparison btw Lar's/my old clone and > > > what's > > > > currently in the Apache/gh repos. > > > > > > > > Notice: > > > > 1) that the only difference is in the name for the author/committer > > name > > > > is missing. > > > > 2) If you look at the actual commit from SVN it includes the user > > name: > > > > https://svn.apache.org/viewvc?view=revision&revision=670801 > > > > > > > > Not sure how this crept in, perhaps a bug in git that was exposed as > > part > > > > of the force push? > > > > > > > > ---- > > > > phunt@phunt-MBP13:~/dev/t/lars[branch-3.3]$ git show --format=raw > > > > 88a823c32dd080ae8a193948ba8915e9fa222e37 > > > > commit 88a823c32dd080ae8a193948ba8915e9fa222e37 > > > > tree 9e2522402b6d26e3c9a6e32ba9fc041f7603a42d > > > > parent 11d2d6fd92acf9abc762c41e0f7b91c5acc89f4f > > > > author mahadevkonar <[email protected]> 1194634898 +0000 > > > > committer mahadevkonar <[email protected]> 1194634898 +0000 > > > > > > > > adding junit jar file for ant tests to run without installing > junit > > > > > > > > git-svn-id: https://svn.apache.org/repos/ > > asf/hadoop/zookeeper/trunk@ > > > > 670801 13f79535-47bb-0310-9956-ffa450edef68 > > > > > > > > diff --git a/zookeeper/java/lib/junit-4.1.jar > > > > b/zookeeper/java/lib/junit-4. > > > > 1.jar > > > > new file mode 100644 > > > > index 00000000..2c46d67a > > > > Binary files /dev/null and b/zookeeper/java/lib/junit-4.1.jar differ > > > > phunt@phunt-MBP13:~/dev/t/lars[branch-3.3]$ > > > > ---- > > > > > > > > phunt@dhcp-10-16-2-37:~/dev/t/zookeeper[branch-3.3]$ git show > > > --format=raw > > > > 1d2cb51508281a281bc47d6bf93fbf264cc20540 > > > > commit 1d2cb51508281a281bc47d6bf93fbf264cc20540 > > > > tree 9e2522402b6d26e3c9a6e32ba9fc041f7603a42d > > > > parent 11d2d6fd92acf9abc762c41e0f7b91c5acc89f4f > > > > author <[email protected]> 1194634898 +0000 > > > > committer <[email protected]> 1194634898 +0000 > > > > > > > > adding junit jar file for ant tests to run without installing > junit > > > > > > > > git-svn-id: https://svn.apache.org/repos/ > > asf/hadoop/zookeeper/trunk@ > > > > 670801 13f79535-47bb-0310-9956-ffa450edef68 > > > > > > > > diff --git a/zookeeper/java/lib/junit-4.1.jar > > > > b/zookeeper/java/lib/junit-4. > > > > 1.jar > > > > new file mode 100644 > > > > index 00000000..2c46d67a > > > > Binary files /dev/null and b/zookeeper/java/lib/junit-4.1.jar differ > > > > phunt@dhcp-10-16-2-37:~/dev/t/zookeeper[branch-3.3]$ > > > > ---- > > > > > > > > Patric. > > > > > > > > On Tue, Sep 12, 2017 at 10:40 AM, Michael Han <[email protected]> > wrote: > > > > > > > > > The tags themselves were not moved or force pushed as far as I can > > > tell. > > > > > The problem is the every commit differs in terms of the commit hash > > > > between > > > > > the official ZooKeeper repo and the old repos (autodesk-forks and > > > phunt). > > > > > > > > > > I suspect it was the svn git move that causes this: > > > > > https://issues.apache.org/jira/browse/INFRA-12573. As part of the > > > move, > > > > > the > > > > > main branch of our repo was renamed from "trunk" to "master". From > > > what I > > > > > know, git generates commit hash based on a couple of parameters > > > including > > > > > the source tree of the commit, and maybe the renaming of the branch > > > here > > > > > changed the state of the source tree that leads to the rewriting of > > the > > > > > entire commit hash history (but not the commit contents, as they > were > > > > > intact.). > > > > > > > > > > I think a reasonable next step is to confirm with someone in INFRA > > > about > > > > > this while waiting for more inputs from this group. > > > > > > > > > > On Tue, Sep 12, 2017 at 9:23 AM, Patrick Hunt <[email protected]> > > > wrote: > > > > > > > > > > > The tags differ between old clones and recent ones. > > > > > > > > > > > > https://git-wip-us.apache.org/repos/asf/zookeeper.git > > > > > > [email protected]:apache/zookeeper.git > > > > > > git://git.apache.org/zookeeper.git > > > > > > > > > > > > are all consistent (tags) as are > > > > > > > > > > > > https://github.com/autodesk-forks/zookeeper.git > > > > > > https://github.com/phunt/zookeeper.git (my old repo) > > > > > > > > > > > > I checked out a few tags from lars repo as well as apache latest, > > > while > > > > > the > > > > > > tags differ the content does not. > > > > > > > > > > > > Using gitk I see that the two sets of repos differ at > > > > > > commit 11d2d6fd92acf9abc762c41e0f7b91c5acc89f4f > > > > > > > > > > > > I'm not sure why (have to run to a meeting...) > > > > > > > > > > > > Patrick > > > > > > > > > > > > > > > > > > On Tue, Sep 12, 2017 at 7:46 AM, Patrick Hunt <[email protected]> > > > > wrote: > > > > > > > > > > > > > Hm.. I do see a relatively recent force push: > > http://markmail.org/ > > > > > > > message/n3ho5n7eb6uvfacs and associated jira > > https://issues.apache > > > . > > > > > > > org/jira/browse/INFRA-13916 > > > > > > > > > > > > > > Michael? > > > > > > > > > > > > > > Patrick > > > > > > > > > > > > > > On Tue, Sep 12, 2017 at 7:31 AM, Patrick Hunt < > [email protected]> > > > > > wrote: > > > > > > > > > > > > > >> I haven't had a chance to look into this, could be serious - > > does > > > > > anyone > > > > > > >> know if this was related to the move from SVN? > > > > > > >> > > > > > > >> Patrick > > > > > > >> > > > > > > >> On Mon, Sep 11, 2017 at 10:29 AM, Lars Schneider < > > > > > > >> [email protected]> wrote: > > > > > > >> > > > > > > >>> Hi, > > > > > > >>> > > > > > > >>> I fetched changes from git://git.apache.org/zookeeper.git > and > > > > > noticed > > > > > > >>> that almost all released tags changed. > > > > > > >>> E.g. my "release-3.5.2" is at 379eb9c and the upstream > > > > > "release-3.5.2" > > > > > > >>> 3f572f0. > > > > > > >>> > > > > > > >>> There is no difference between the tags: > > > > > > >>> $ git diff 379eb9c..3f572f0 | wc -l > > > > > > >>> 0 > > > > > > >>> > > > > > > >>> I am reasonably sure that I did not change those tags. > > > > > > >>> Do you know what happened? I wasn't able to find anything > about > > > > this > > > > > in > > > > > > >>> the mailing list archive. > > > > > > >>> > > > > > > >>> Thanks, > > > > > > >>> Lars > > > > > > >>> > > > > > > >>> > > > > > > >>> Here is the full list of different tags: > > > > > > >>> > > > > > > >>> 0.0.1 > > > > > > >>> 2.1.0 > > > > > > >>> apache-initial-import > > > > > > >>> post_ZOOKEEPER-45_snapshot > > > > > > >>> release-3.0.0 > > > > > > >>> release-3.0.1 > > > > > > >>> release-3.0.1-rc0 > > > > > > >>> release-3.1.0 > > > > > > >>> release-3.1.0-rc0 > > > > > > >>> release-3.1.0-rc1 > > > > > > >>> release-3.1.0-rc2 > > > > > > >>> release-3.1.1 > > > > > > >>> release-3.1.1-rc0 > > > > > > >>> release-3.1.2 > > > > > > >>> release-3.1.2-rc0 > > > > > > >>> release-3.2.0 > > > > > > >>> release-3.2.0-rc0 > > > > > > >>> release-3.2.1 > > > > > > >>> release-3.2.2 > > > > > > >>> release-3.2.2-rc0 > > > > > > >>> release-3.3.0 > > > > > > >>> release-3.3.1 > > > > > > >>> release-3.3.2 > > > > > > >>> release-3.3.3 > > > > > > >>> release-3.3.3-rc0 > > > > > > >>> release-3.3.3-rc1 > > > > > > >>> release-3.3.4 > > > > > > >>> release-3.3.4-rc0 > > > > > > >>> release-3.3.5 > > > > > > >>> release-3.3.5-rc0 > > > > > > >>> release-3.3.5-rc1 > > > > > > >>> release-3.3.6 > > > > > > >>> release-3.3.6-rc0 > > > > > > >>> release-3.4.0 > > > > > > >>> release-3.4.0-rc0 > > > > > > >>> release-3.4.0-rc1 > > > > > > >>> release-3.4.0-rc2 > > > > > > >>> release-3.4.1 > > > > > > >>> release-3.4.1-rc0 > > > > > > >>> release-3.4.2 > > > > > > >>> release-3.4.2-rc0 > > > > > > >>> release-3.4.3 > > > > > > >>> release-3.4.3-rc0 > > > > > > >>> release-3.4.4 > > > > > > >>> release-3.4.4-rc0 > > > > > > >>> release-3.4.5 > > > > > > >>> release-3.4.5-rc0 > > > > > > >>> release-3.4.5-rc1 > > > > > > >>> release-3.4.6 > > > > > > >>> release-3.4.6-rc0 > > > > > > >>> release-3.4.7 > > > > > > >>> release-3.4.7-rc0 > > > > > > >>> release-3.4.8 > > > > > > >>> release-3.4.8-rc0 > > > > > > >>> release-3.4.9-rc0 > > > > > > >>> release-3.4.9-rc1 > > > > > > >>> release-3.4.9-rc2 > > > > > > >>> release-3.5.0 > > > > > > >>> release-3.5.0-rc0 > > > > > > >>> release-3.5.1 > > > > > > >>> release-3.5.1-rc0 > > > > > > >>> release-3.5.1-rc1 > > > > > > >>> release-3.5.1-rc2 > > > > > > >>> release-3.5.1-rc3 > > > > > > >>> release-3.5.1-rc4 > > > > > > >>> release-3.5.2 > > > > > > >>> release-3.5.2-rc0 > > > > > > >>> release-3.5.2-rc1 > > > > > > >> > > > > > > >> > > > > > > >> > > > > > > > > > > > > > > > > > > > > > > > > > > > >
