About half way done with implementing pairing. The webapp's interface to prompt for a secret and start pairing is done; the protocol is implemented; broadcasting of pairing requests is working; added Yet Another Thread to listen for incoming pairing traffic.

Very happy with how this came together; starting with defining the protocol with data types let me rapidly iterate until I had designed a simple, clean, robust protocol. The implementation works well too; it's even possible to start pairing, and only then bring up the network interface to the machine you intended to pair with, and it'll detect the new interface and start sending requests to it.

Next, I need to make alerts have a button that performs a stored IO action. So that the incoming pair request alert can have a button to respond to the pair request. And then I need to write the code to actually perform the pairing, including ssh key setup.

How do you prefer bugs to be reported?

I'm having a problem where git-annex (from git, 1e41c0d85ecc24e8656bff79b2fba46c3663a054) is taking over 20GB of RAM after adding a single file.

To reproduce:

  1. Create a new annex (I used the web app, created an annex at ~/annex)
  2. Add box.com as a remote, using encryption (followed http://git-annex.branchable.com/tips/using_box.com_as_a_special_remote/, basically cd ~/annex; git annex initremote box.com type=directory directory=/media/box.com encryption=$GPGID)
  3. copied a file into the annex (ok, I admit, the file was a bit large, 350MB, but still)
  4. refreshed the web app, nothing happened
  5. closed the web app
  6. started git annex assistant inside ~/annex
  7. git annex webapp
  8. noted that git annex noticed the file, and started transferring
  9. wait
  10. memory usage for the git-annex process goes beyond 21GB RAM
  11. oomkiller kills git-annex

The bugs page is the place to put bug reports like this so I won't forget them.

This should certianly not be happening. There are actually two git-annex processes running in the situation you describe; I'd be most curious to know whether the git annex transfer process was the one that blew up, or if the git annex assistant blew up. Also, it's not clear to me if you enabled the chunksize parameter when setting up the special remote, which could well be a significant detail.

Comment by http://joeyh.name/ Sun Sep 9 16:25:52 2012
I'm almost certain that I specified chunksize like in the tip. How do I check?

You can see the full configuration with: git show git-annex:remote.log

(You probably do not want to paste the cipher= part of that here, although it is encrypted with your gpg key, unless you are using the shared encryption mode.)

Comment by http://joeyh.name/ Sun Sep 9 18:00:48 2012
chunksize=2m was defined. I tried again, and the transfers again didn't seem to start. Then I did killall git-annex, cleaned the files from the box.com account, and tried again. This time it seems to be working, but the webapp isn't showing progress. Memory usage also stays reasonable, and doesn't seem to grow. Maybe it was just bad handling of unexpected repository state?

Re transfer progress not being shown, that's expected as upload progress displays are not yet implemented in the webapp.

I can't think of a way git-annex would care what was in your box.com repo. It just makes directories as needed, and will overwrite existing files in the rare case they already exist. And ignore any non-git-annex files.

Are you sure it was git-annex's memory use blowing up, and not the memory use of the davfs2 daemon?

Comment by http://joeyh.name/ Mon Sep 10 17:31:20 2012
Positive. I was grepping for git-annex from ps for the mem usage.
Comments on this page are closed.