P3D3::MacHG temporarily shows corruption message.

Issue #190 open
Tony Hartmann
created an issue

Hi,

it has happened to me several times now that machg destroys a repository (not the files though) when treating a remote repo (mounted as a disk eg. via forklift) as a local folder.

The practice itself works like a charm because now only ftp is needed to also work with hg repos on remote servers ... but machg does not seem to like it.

The Bugs i can report do not happen immediately, machg actually works as supposed to for a couple minutes/commits but then suddenly it goes nuts.

MacHG returns the error message:

Error During Outgoing - Mercurial reported error number 255: abort: index 00chnagelog.i is corrupted!

So basically it makes the repository unusable.

The second issue i noted is that MacHg eats lots of resources (do not know if that is also linked to the bug) with lots of resources i mean, it makes my fan on the MacBook Pro run wild. Looking at the the Activity Monitor MacHG is eating more than 100% of the CPU. There is quite some threads open - also more than one 100 and a lot of RAM is used as well.

MacHG is a great app and i would not hesitate spending a buck or two for getting this issues fixed.

Thanks in advance T.H.

Comments (8)

  1. Jason Harris repo owner

    First thanks for your report!

    Second, can I get you to file the first part of this issue again and explain in more detail exactly what is going on? Is there any way you can reproduce this each and every time.

    There is one circumstance when you are changing the repository out from under MacHg. Ie replacing parts of it via ftp or a finder copy. MacHg detects that something has changed in the repository and queries the repository for the new changes. However at that very moment the repository might be only partially complete ie only part of the change has completed, then the repository is in a state which appears "corrupted". Of course once the copy finishes then you are back to a state where the repository is "whole" again. Ie its working fine again. This only happens as well when you copy over the repository/.hg directory outside normal Mercurial operations. (eg via ftp or finder copy or something...)

    Please test to see that your repository is in fact still corrupted after the operation you are doing completes. My strong guess is that MacHg has just caught the repository in the middle of some external operation...

    (if you are doing the updating via push or pull then Mercurial has locking mechanisms which stop you accessing the repository while it is changing...)

    Does this sound like your issue?

    As for the second part that was reported as #148 and is now fixed in development a new MacHg will be issued shortly.

  2. Jason Harris repo owner

    Also could I get you to get a snapshot of the alert message that comes up? I want to use it in a FAQ item and its a bit hard to produce artificially...

    Thanks! Jas

  3. Tony Hartmann reporter

    Hello Jason,

    thanks for your quick response. I like your attitude you have towards your project and the users that use it. Way to go!

    Unfortunately the issue i reported here is not yet resolved. :-(

    I can actually reproduce the issue each time. I have attached a couple screen-shots this time, so that you can analyze this thing better.

    Basically here is the steps: 1.) Fire up ForkLift to mount remote FTP into local filesystem. 2.) Remote Rep - is now seen by PC/Mac as local rep as it can be accessed via Finder, Terminal etc. 3.) Fire up MacHG to get changes from bitbucket server. 4.) MacHG pulls changes and updates local dev repository just fine. 5.) Pushing changes to mounted remote repository (mounted in step1). -> push is being reported by mag hg as successful push 6.) Updating Remote Mounted Repository -> Error (screen-shot 1) 7.) After error repository does not work anymore (screen-shot 2) 8.) MacHG start consuming a lot of CPU Resource (screen-shot 3). Fans start going wild until MacHG is shut down. If machg is not shut down - resources being consumed add up quickly.

    Ok hope i could give some insights into the issue. As noted - i would be really thankful if this error can be fixed in some way. MacHG is a pretty cool App and once all bugs are gone - it will rock even more :-)

    cheers tony

  4. Tony Hartmann reporter

    Update:

    Just happened to me again. But when checking it closely, i actually think the push that is reported to be successful is already not correct. Because after the push MacHG does not update the current working head of the remote mounted repository. All it shows the is the current head of the local development version.

    So i think not the update but the push command is screwing up the repository.

  5. Tony Hartmann reporter

    Hi Jason,

    i was using version 0.9.13 - updated just now. I will check if the cpu consuming habit is gone.

    cheers tony

    btw. your donation thank you page is unavailable ;-)

  6. Log in to comment