projectbismark / bismark-passive

This repository does not currently contain any code of its own -- it is a container for project-wide issues and wiki pages to be stored. Code is stored in many of the bismark-* repositories.

Wiki: https://github.com/projectbismark/projectbismark/wiki

Project-wide issues: https://github.com/projectbismark/projectbismark/issues

Overall issue dashboard: https://github.com/organizations/projectbismark/dashboard/issues

Project Page and Dataset: http://projectbismark.net.s3.us-east-2.amazonaws.com/index.html

bismark-passive's Issues

Use syslog

Backup encrypted ANONYMIZATION_SEED_FILE on server to survive reflashes

Continuity of bismark-passive data will be lost if a bismark-passive user upgrades or reflashes their router. It is necessary to save and restore the ANONYMIZATION_SEED_FILE to ensure data continuity. This process is not currently automated -- it would be incumbent on participants to perform the save and restore themselves.

These files could be backed up to a Georgia Tech server if they were encrypted by a key generated from a user-provided password using something like bcrypt or scrypt (http://www.tarsnap.com/scrypt.html). Thus, as long as we enforced some minimum password quality requirement, it would be possible for us to store these files (and other conf files) while being unable to access their contents.

Files occasionally get tarred twice

For example, OWC43DC78EE081-1338829308451006-119723.gz in OWC43DC78EE081_2012-07-16_08-15-10.tar is actually a tar file containing several gzipped updates itself (not including 119723, so there's a file missing.)

Seems to happen when there's a backlog of data on the router. In this case, there was an hour of data in the tarfile. The corrupted gzip was the first sequence number in the tarfile.

Might be a race condition with staging the updates on either the client or server?

Restart if data is lost

We should detect when bismark-data-transmit cannot upload files and restart bismark-passive when that happens.