diff options
Diffstat (limited to 'doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies')
6 files changed, 95 insertions, 0 deletions
diff --git a/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_1_b23c42004a3486d2409c1f96afa819aa._comment b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_1_b23c42004a3486d2409c1f96afa819aa._comment new file mode 100644 index 000000000..c059040e1 --- /dev/null +++ b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_1_b23c42004a3486d2409c1f96afa819aa._comment @@ -0,0 +1,18 @@ +[[!comment format=mdwn + username="emanuele.olivetti@47d88ed185b03191e25329caa6fabc2efb3118b2" + nickname="emanuele.olivetti" + subject="comment 1" + date="2016-02-13T22:27:56Z" + content=""" +Additional information: I have 62 files in the added directory. Git-annex info return this, with a suspicious \"numcopies -2: 62\": + + > git-annex info . + directory: . + local annex keys: 62 + local annex size: 377.19 megabytes + annexed files in working tree: 62 + size of annexed files in working tree: 377.19 megabytes + numcopies stats: + numcopies -2: 62 + +"""]] diff --git a/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_2_e379c25d18ee1ae4ea7e0a5a33e1c75e._comment b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_2_e379c25d18ee1ae4ea7e0a5a33e1c75e._comment new file mode 100644 index 000000000..c9e046aa6 --- /dev/null +++ b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_2_e379c25d18ee1ae4ea7e0a5a33e1c75e._comment @@ -0,0 +1,11 @@ +[[!comment format=mdwn + username="joey" + subject="""comment 2""" + date="2016-02-14T19:13:05Z" + content=""" +Is the content of the files locally present? Either fsck incorrectly thinks +it's not present despite it being present, or fsck correctly noticed that +it somehow went missing.. + +Is this repository using direct mode? +"""]] diff --git a/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_3_063e27ec1f2dd23fbf914a08213316df._comment b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_3_063e27ec1f2dd23fbf914a08213316df._comment new file mode 100644 index 000000000..6146fad5c --- /dev/null +++ b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_3_063e27ec1f2dd23fbf914a08213316df._comment @@ -0,0 +1,12 @@ +[[!comment format=mdwn + username="emanuele.olivetti@47d88ed185b03191e25329caa6fabc2efb3118b2" + nickname="emanuele.olivetti" + subject="comment 3" + date="2016-02-15T09:47:24Z" + content=""" +Yes, the content is locally present. In the sense that each symlink in the current directory points to an actual file in .git/annex/objects/ . That is why, in my opinion, fsck incorrectly thinks the content is not present, despite it being present. + +Initially I thought that a possible explanation was that 'git commit' went wrong, so that I would have to do it again. I tried to git commit again, but git says there is nothing to commit. + +My repository is in indirect mode. All files in the repository, including the troubling ones are in .git/annex/objects. +"""]] diff --git a/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_4_8688477bb694dbc9e6c7768f5f375f3f._comment b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_4_8688477bb694dbc9e6c7768f5f375f3f._comment new file mode 100644 index 000000000..1fa409f9c --- /dev/null +++ b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_4_8688477bb694dbc9e6c7768f5f375f3f._comment @@ -0,0 +1,28 @@ +[[!comment format=mdwn + username="emanuele.olivetti@47d88ed185b03191e25329caa6fabc2efb3118b2" + nickname="emanuele.olivetti" + subject="repository marked as dead (!)" + date="2016-02-18T14:33:56Z" + content=""" +Update: I've just tried to add a 'test' directory to the repository with only a 'foo' file inside, then git commit. Surprisingly, if I do 'git annex fsck .' of it, the repository is marked as **dead** and no known copies of foo are available: + + (master)> mkdir test ; cd test + test (master) > cat > foo + bar + test (master)> git annex add . + add foo ok + (recording state in git...) + test (master)> git commit -m \"added fake content to test git annex repo\" + [master b9f0a8f] added fake content to test git annex repo + 1 file changed, 1 insertion(+) + create mode 120000 events/2015/test/foo + test (master)> git annex fsck . + Warning: Fscking a repository that is currently marked as dead. + fsck foo (checksum...) + ** No known copies exist of foo + failed + (recording state in git...) + git-annex: fsck: 1 failed + +At this point, I suspect that the repository on my laptop has some serious problem. Shall I move this discussion to the forum? It may be a git-annex bug but it can just be that the repository is damaged for other reasons. Moreover, what is the best course of action, at this point? +"""]] diff --git a/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_5_a256583bd9b3815a23cd1ca40d6c19ca._comment b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_5_a256583bd9b3815a23cd1ca40d6c19ca._comment new file mode 100644 index 000000000..785d6c7b7 --- /dev/null +++ b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_5_a256583bd9b3815a23cd1ca40d6c19ca._comment @@ -0,0 +1,10 @@ +[[!comment format=mdwn + username="emanuele.olivetti@47d88ed185b03191e25329caa6fabc2efb3118b2" + nickname="emanuele.olivetti" + subject="'git-annex semitrust' seems to solve the issue" + date="2016-02-18T21:51:00Z" + content=""" +Given that the repository is marked as dead, I tried to bring it back with 'git-annex semitrust laptop' - 'laptop' being the name of the git annex repository on my laptop. Apparently, this action solved the problem. Fsck now works as expected, and I could sync and get the new files on the clone on the external hard disk as usual. + +Now, if this the correct solution to my problem, I am wondering how it is possible that the repository was marked as dead. I carefully checked the history of previous command, but no sign of 'git annex dead' o similar. Are there scenarios where a repository is automatically marked as dead? +"""]] diff --git a/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_6_0999f9ec9c3d6f51889141344d4cfcb6._comment b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_6_0999f9ec9c3d6f51889141344d4cfcb6._comment new file mode 100644 index 000000000..a2e678c1b --- /dev/null +++ b/doc/bugs/after_git_annex_add_and_commit__44___git_annex_fsck_fails__58___no_known_copies/comment_6_0999f9ec9c3d6f51889141344d4cfcb6._comment @@ -0,0 +1,16 @@ +[[!comment format=mdwn + username="joey" + subject="""comment 6""" + date="2016-02-19T18:52:25Z" + content=""" +Ok, the repository being dead certianly explains why fsck found a problem. + +It may be that fsck should have a better message in this case. Like +"only dead repository had copies of the file". + +Repositories are only ever marked dead on user request. There are a few +ways besides `git annex dead` that you could have done it. Perhaps trying +to delete the repo in the webapp, or changing it in `git annex vicfg`. +You can check out the git-annex branch and examine the history of the +trust.log file to see when it was changed to dead. +"""]] |