git-annex.git
10 years agoclean up side lock files when we're done with them
Joey Hess [Mon, 16 Nov 2015 15:36:11 +0000 (11:36 -0400)]
clean up side lock files when we're done with them

There's a potential race, but it's detected and just results in the other
process failing to take the side lock, so possibly retrying one second
later on. The race window is quite narrow so the extra delay is minor.

Left the side lock files mode 666 because an interruption can leave a side
lock file created by another user for a shared repository. When this
happens, the non-owning user can't delete it (+t) but can still lock it,
and so the code falls back to acting as it did before this commit.

10 years agofix typo
edward [Sat, 14 Nov 2015 09:57:40 +0000 (09:57 +0000)]
fix typo

10 years agodevblog
Joey Hess [Fri, 13 Nov 2015 20:34:54 +0000 (16:34 -0400)]
devblog

10 years agoclose
Joey Hess [Fri, 13 Nov 2015 20:33:53 +0000 (16:33 -0400)]
close

10 years agostarting to get a handle on how to detect that mad gleam in lustre's eye
Joey Hess [Fri, 13 Nov 2015 20:13:43 +0000 (16:13 -0400)]
starting to get a handle on how to detect that mad gleam in lustre's eye

10 years agocomment
Joey Hess [Fri, 13 Nov 2015 20:13:38 +0000 (16:13 -0400)]
comment

10 years agomore warnings about networked filesystems
Joey Hess [Fri, 13 Nov 2015 19:55:16 +0000 (15:55 -0400)]
more warnings about networked filesystems

10 years agoone more try to get sane behavior our of lustre
Joey Hess [Fri, 13 Nov 2015 19:51:45 +0000 (15:51 -0400)]
one more try to get sane behavior our of lustre

10 years agofixed a fd double-close
Joey Hess [Fri, 13 Nov 2015 19:43:09 +0000 (15:43 -0400)]
fixed a fd double-close

10 years agoalso compare lock file contents to double-check link worked
Joey Hess [Fri, 13 Nov 2015 19:20:52 +0000 (15:20 -0400)]
also compare lock file contents to double-check link worked

And it closes the tmp file before this. I don't know if this will help
avoid lustre's craziness, but it can't hurt..

10 years agouse /tmp for sidelock file when no /dev/shm
Joey Hess [Fri, 13 Nov 2015 18:49:30 +0000 (14:49 -0400)]
use /tmp for sidelock file when no /dev/shm

10 years agorequire the side lock be held to take pidlock
Joey Hess [Fri, 13 Nov 2015 18:44:53 +0000 (14:44 -0400)]
require the side lock be held to take pidlock

This is less portable, since currently sidelocks rely on /dev/shm.
But, I've seen crazy lustre inconsistencies that make me not trust the
link() method at all, so what can you do.

10 years agoavoid over-long filenames for side lock files
Joey Hess [Fri, 13 Nov 2015 18:04:29 +0000 (14:04 -0400)]
avoid over-long filenames for side lock files

10 years agocomment
Joey Hess [Fri, 13 Nov 2015 17:47:14 +0000 (13:47 -0400)]
comment

10 years agoupdate
Joey Hess [Fri, 13 Nov 2015 17:46:11 +0000 (13:46 -0400)]
update

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Fri, 13 Nov 2015 17:35:46 +0000 (13:35 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agoinit: Automatically enable annex.pidlock when necessary.
Joey Hess [Fri, 13 Nov 2015 17:35:29 +0000 (13:35 -0400)]
init: Automatically enable annex.pidlock when necessary.

10 years agoadd stat check
Joey Hess [Fri, 13 Nov 2015 17:22:45 +0000 (13:22 -0400)]
add stat check

I have a strace taken on a lustre filesystem on which link() returned 0,
but didn't actually succeed, since the file already existed.

One of the linux man pages recommended using link followed by checking like
this. I was reading it yesterday, but cannot find it now.

10 years agoclean up temp file
Joey Hess [Fri, 13 Nov 2015 16:52:24 +0000 (12:52 -0400)]
clean up temp file

10 years agobetter lock dropping order
Joey Hess [Fri, 13 Nov 2015 16:36:37 +0000 (12:36 -0400)]
better lock dropping order

10 years agoAdded a comment: PLEASE!!! PLEASE BUILD IT!!!
SchemaCzar [Fri, 13 Nov 2015 16:10:44 +0000 (16:10 +0000)]
Added a comment: PLEASE!!! PLEASE BUILD IT!!!

10 years agoWorkflow question
juh [Fri, 13 Nov 2015 13:41:03 +0000 (13:41 +0000)]
Workflow question

10 years ago(no commit message)
atrent [Fri, 13 Nov 2015 07:39:42 +0000 (07:39 +0000)]

10 years ago(no commit message)
atrent [Fri, 13 Nov 2015 07:38:01 +0000 (07:38 +0000)]

10 years ago(no commit message)
atrent [Fri, 13 Nov 2015 07:37:20 +0000 (07:37 +0000)]

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Thu, 12 Nov 2015 22:24:56 +0000 (18:24 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agodevblog
Joey Hess [Thu, 12 Nov 2015 22:24:26 +0000 (18:24 -0400)]
devblog

10 years agoconvert from Utility.LockPool to Annex.LockPool everywhere
Joey Hess [Thu, 12 Nov 2015 22:05:45 +0000 (18:05 -0400)]
convert from Utility.LockPool to Annex.LockPool everywhere

10 years agogeneralize to MonadIO
Joey Hess [Thu, 12 Nov 2015 22:03:49 +0000 (18:03 -0400)]
generalize to MonadIO

10 years agopid locking configuration and abstraction layer for git-annex
Joey Hess [Thu, 12 Nov 2015 21:47:31 +0000 (17:47 -0400)]
pid locking configuration and abstraction layer for git-annex

(not actually used anywhere yet)

10 years agoadd timeout for pid lock waiting
Joey Hess [Thu, 12 Nov 2015 21:12:54 +0000 (17:12 -0400)]
add timeout for pid lock waiting

10 years agorefactor
Joey Hess [Thu, 12 Nov 2015 20:35:15 +0000 (16:35 -0400)]
refactor

10 years agomodule for PidLocks in LockPool
Joey Hess [Thu, 12 Nov 2015 20:31:34 +0000 (16:31 -0400)]
module for PidLocks in LockPool

10 years agomake LockPool's LockHandle be able to support multiple different types of file locks
Joey Hess [Thu, 12 Nov 2015 20:28:11 +0000 (16:28 -0400)]
make LockPool's LockHandle be able to support multiple different types of file locks

10 years agomodule for pid lock files with atomic stale lock file takeover when possible
Joey Hess [Thu, 12 Nov 2015 19:38:02 +0000 (15:38 -0400)]
module for pid lock files with atomic stale lock file takeover when possible

10 years agoinitial complaint
https://me.yahoo.com/a/EbvxpTI_xP9Aod7Mg4cwGhgjrCrdM5s-#7c0f4 [Thu, 12 Nov 2015 17:09:38 +0000 (17:09 +0000)]
initial complaint

10 years agocreate ipfs bug.
fiatjaf [Thu, 12 Nov 2015 12:20:52 +0000 (12:20 +0000)]
create ipfs bug.

10 years agoAdded a comment: obnam
https://woid.cryptobitch.de/foobar [Thu, 12 Nov 2015 11:50:52 +0000 (11:50 +0000)]
Added a comment: obnam

10 years agoAdded a comment
https://me.yahoo.com/a/EbvxpTI_xP9Aod7Mg4cwGhgjrCrdM5s-#7c0f4 [Wed, 11 Nov 2015 19:28:41 +0000 (19:28 +0000)]
Added a comment

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 11 Nov 2015 19:08:57 +0000 (15:08 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agodesign
Joey Hess [Wed, 11 Nov 2015 18:26:00 +0000 (14:26 -0400)]
design

10 years agoAdded a comment
alejandro@2e9e229fdf45d44cd300a5681432552693d458ab [Wed, 11 Nov 2015 15:22:06 +0000 (15:22 +0000)]
Added a comment

10 years ago(no commit message)
JulianMarsalis [Wed, 11 Nov 2015 15:14:44 +0000 (15:14 +0000)]

10 years agoAdded a comment
https://id.koumbit.net/anarcat [Wed, 11 Nov 2015 14:53:12 +0000 (14:53 +0000)]
Added a comment

10 years agoAdded a comment
alejandro@2e9e229fdf45d44cd300a5681432552693d458ab [Wed, 11 Nov 2015 08:29:09 +0000 (08:29 +0000)]
Added a comment

10 years agoAdded a comment: why use rsync at all?
https://id.koumbit.net/anarcat [Wed, 11 Nov 2015 01:13:01 +0000 (01:13 +0000)]
Added a comment: why use rsync at all?

10 years ago(no commit message)
alejandro@2e9e229fdf45d44cd300a5681432552693d458ab [Tue, 10 Nov 2015 23:46:10 +0000 (23:46 +0000)]

10 years agodevblog
Joey Hess [Tue, 10 Nov 2015 21:11:44 +0000 (17:11 -0400)]
devblog

10 years agoI'm feeling.. typoish
Joey Hess [Tue, 10 Nov 2015 21:00:06 +0000 (17:00 -0400)]
I'm feeling.. typoish

10 years agofix
Joey Hess [Tue, 10 Nov 2015 20:59:12 +0000 (16:59 -0400)]
fix

10 years agouse author()
Joey Hess [Tue, 10 Nov 2015 20:57:48 +0000 (16:57 -0400)]
use author()

10 years agoadd userpage for yoh
Joey Hess [Tue, 10 Nov 2015 20:55:53 +0000 (16:55 -0400)]
add userpage for yoh

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 10 Nov 2015 20:53:06 +0000 (16:53 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agoassistant: Pass ssh-options through 3 more git pull/push calls that were missed before.
Joey Hess [Tue, 10 Nov 2015 20:52:30 +0000 (16:52 -0400)]
assistant: Pass ssh-options through 3 more git pull/push calls that were missed before.

It was used for regular pull, but not for regular push, tagged push, or the
fallback fetching.

10 years agoremoved
aloukian [Tue, 10 Nov 2015 20:36:45 +0000 (20:36 +0000)]
removed

10 years agoAdded a comment
aloukian [Tue, 10 Nov 2015 20:36:17 +0000 (20:36 +0000)]
Added a comment

10 years agoAdded a comment
aloukian [Tue, 10 Nov 2015 20:35:59 +0000 (20:35 +0000)]
Added a comment

10 years agoresponse
Joey Hess [Tue, 10 Nov 2015 19:52:12 +0000 (15:52 -0400)]
response

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 10 Nov 2015 19:38:58 +0000 (15:38 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agocomment
Joey Hess [Tue, 10 Nov 2015 19:38:33 +0000 (15:38 -0400)]
comment

10 years agoadded [[!meta author=yoh]]
https://me.yahoo.com/a/EbvxpTI_xP9Aod7Mg4cwGhgjrCrdM5s-#7c0f4 [Tue, 10 Nov 2015 19:34:46 +0000 (19:34 +0000)]
added [[!meta author=yoh]]

10 years agoresponse
Joey Hess [Tue, 10 Nov 2015 19:29:10 +0000 (15:29 -0400)]
response

10 years agoadded [[!meta author=yoh]]
https://me.yahoo.com/a/EbvxpTI_xP9Aod7Mg4cwGhgjrCrdM5s-#7c0f4 [Tue, 10 Nov 2015 19:25:28 +0000 (19:25 +0000)]
added [[!meta author=yoh]]

10 years agoAdded a comment
https://me.yahoo.com/a/EbvxpTI_xP9Aod7Mg4cwGhgjrCrdM5s-#7c0f4 [Tue, 10 Nov 2015 19:24:59 +0000 (19:24 +0000)]
Added a comment

10 years agoresponse
Joey Hess [Tue, 10 Nov 2015 19:16:43 +0000 (15:16 -0400)]
response

10 years agohmm
Joey Hess [Tue, 10 Nov 2015 19:12:12 +0000 (15:12 -0400)]
hmm

10 years agoresponse
Joey Hess [Tue, 10 Nov 2015 19:03:10 +0000 (15:03 -0400)]
response

10 years agoresponse
Joey Hess [Tue, 10 Nov 2015 18:58:38 +0000 (14:58 -0400)]
response

10 years agoclarify
Joey Hess [Tue, 10 Nov 2015 18:49:34 +0000 (14:49 -0400)]
clarify

10 years agodisplay a message in the unlikely scenario of fsking a dead repository
Joey Hess [Tue, 10 Nov 2015 18:44:58 +0000 (14:44 -0400)]
display a message in the unlikely scenario of fsking a dead repository

10 years agoadd back missing newline to showRaw
Joey Hess [Tue, 10 Nov 2015 18:06:07 +0000 (14:06 -0400)]
add back missing newline to showRaw

10 years agofsck: When fscking a dead repo, avoid incorrect "fixing location log" message.
Joey Hess [Tue, 10 Nov 2015 17:59:04 +0000 (13:59 -0400)]
fsck: When fscking a dead repo, avoid incorrect "fixing location log" message.

keyLocations doesn't return locations in dead repos, but if we're fscking a
dead repo, we want to look at what locations are actually logged for it.

10 years agocomment
Joey Hess [Tue, 10 Nov 2015 17:48:15 +0000 (13:48 -0400)]
comment

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 10 Nov 2015 17:44:25 +0000 (13:44 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agofix build warning when built w/o concurrent-output
Joey Hess [Tue, 10 Nov 2015 17:42:39 +0000 (13:42 -0400)]
fix build warning when built w/o concurrent-output

10 years agoAdded a comment
https://me.yahoo.com/a/EbvxpTI_xP9Aod7Mg4cwGhgjrCrdM5s-#7c0f4 [Tue, 10 Nov 2015 17:24:28 +0000 (17:24 +0000)]
Added a comment

10 years agoMake git-annex-standalone.deb include the git-annex html documentation, desktop file...
Joey Hess [Tue, 10 Nov 2015 17:18:58 +0000 (13:18 -0400)]
Make git-annex-standalone.deb include the git-annex html documentation, desktop file, and base completion file, same as the regular git-annex.deb.

It already had a doc-base file relating to the html documentation, and
there's no reason not to include the other stuff.

10 years agoRevert deleton of direct mode page
Joey Hess [Tue, 10 Nov 2015 16:55:41 +0000 (12:55 -0400)]
Revert deleton of direct mode page

This reverts commit 46134e04f8c57da6b02892fb6a9fab05cef5568a.

This may have been an accident, but if it happens again, MARCAO8 will be
banned.

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 10 Nov 2015 16:55:00 +0000 (12:55 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agocomment
Joey Hess [Tue, 10 Nov 2015 16:53:58 +0000 (12:53 -0400)]
comment

10 years agoMake the git-annex-standalone.deb prevent runshell from installing wrappers into...
Joey Hess [Tue, 10 Nov 2015 16:53:46 +0000 (12:53 -0400)]
Make the git-annex-standalone.deb prevent runshell from installing wrappers into $HOME/.ssh

10 years agorunshell: Avoid failing when $HOME/.ssh does not exist and cannot be created.
Joey Hess [Tue, 10 Nov 2015 15:42:06 +0000 (11:42 -0400)]
runshell: Avoid failing when $HOME/.ssh does not exist and cannot be created.

10 years agoresponse
Joey Hess [Tue, 10 Nov 2015 15:32:28 +0000 (11:32 -0400)]
response

10 years ago(no commit message)
MARCAO8 [Tue, 10 Nov 2015 11:49:17 +0000 (11:49 +0000)]

10 years agoclose
Joey Hess [Mon, 9 Nov 2015 16:38:30 +0000 (12:38 -0400)]
close

10 years agoquvi may output utf-8 encoded data when the conifigured locale doesn't support that...
Joey Hess [Mon, 9 Nov 2015 16:19:10 +0000 (12:19 -0400)]
quvi may output utf-8 encoded data when the conifigured locale doesn't support that; avoid crashing on such invalid encoding.

10 years agoclose
Joey Hess [Mon, 9 Nov 2015 15:56:56 +0000 (11:56 -0400)]
close

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 9 Nov 2015 15:34:30 +0000 (11:34 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agoAdded a comment: More details
OlivierBerger [Mon, 9 Nov 2015 14:52:24 +0000 (14:52 +0000)]
Added a comment: More details

10 years agoremove badly named bug report, which was later posted under good name
Joey Hess [Sun, 8 Nov 2015 16:18:00 +0000 (12:18 -0400)]
remove badly named bug report, which was later posted under good name

10 years agoAdded a comment: could / will this help with global status?
https://id.koumbit.net/anarcat [Sat, 7 Nov 2015 16:43:42 +0000 (16:43 +0000)]
Added a comment: could / will this help with global status?

10 years agoAdded a comment
http://www.joachim-breitner.de/ [Sat, 7 Nov 2015 09:02:08 +0000 (09:02 +0000)]
Added a comment

10 years agodevblog
Joey Hess [Fri, 6 Nov 2015 20:03:02 +0000 (16:03 -0400)]
devblog

10 years agowork around git check-ignore --batch bad exit status bug, and bring back import -J
Joey Hess [Fri, 6 Nov 2015 19:39:51 +0000 (15:39 -0400)]
work around git check-ignore --batch bad exit status bug, and bring back import -J

10 years agoadd: Fix error recovery rollback to not move the injested file content out of the...
Joey Hess [Fri, 6 Nov 2015 19:28:20 +0000 (15:28 -0400)]
add: Fix error recovery rollback to not move the injested file content out of the annex back to the file, because other files may point to that same content. Instead, copy the injected file content out to recover.

That was not a data loss, but it came close!

10 years agoenable ConcurrentOutput flag for stack
Joey Hess [Fri, 6 Nov 2015 19:26:12 +0000 (15:26 -0400)]
enable ConcurrentOutput flag for stack

10 years agofix replaceFile makeAnnexLink race
Joey Hess [Fri, 6 Nov 2015 19:08:19 +0000 (15:08 -0400)]
fix replaceFile makeAnnexLink race

replaceFile created a temp file, which was guaranteed to not overlap with
another temp file. However, makeAnnexLink then deleted that file, in
preparation for making the symlink in its place. This caused a race, since
some other replaceFile could create a temp file, using the same name!

I was able to reproduce the race easily running git-annex add -J10 in a
directory with 100 files (all with different contents). Some files would
get ingested into the annex, but their annex links would fail to be added.

There could be other situations where this same problem could occur.
Perhaps when the assistant is adding a file, if the user manually also ran
git-annex add. Perhaps in cases not involving adding a file.

The new replaceFile makes a temprary directory, which is guaranteed to be
unique, and doesn't make a temp file in there. makeAnnexLink can thus
create the symlink without problem and the race is avoided.

Audited all calls to replaceFile to make sure that the old behavior of
providing an empty temp file was not relied on.

The general problem of asking for a temp file and deleting it as part of
the process of using it could reach beyond replaceFile. Did some quick
audits and didn't find other cases of it. Probably only symlink creation
stuff would tend to make that mistake, mostly.

10 years agoclose
Joey Hess [Fri, 6 Nov 2015 17:52:47 +0000 (13:52 -0400)]
close

10 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Fri, 6 Nov 2015 17:45:27 +0000 (13:45 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

10 years agoConcurrent progress bars are now displayed when using -J with a command that moves...
Joey Hess [Fri, 6 Nov 2015 17:44:57 +0000 (13:44 -0400)]
Concurrent progress bars are now displayed when using -J with a command that moves file contents around.