It seems trac is being hit by a high level of automated crawling, to the
extent that use by humans is often slow or entirely unresponsive.
To address this, I've removed the permissions from anonymous access
which are the ones which cause it to fetch data from git (or at least
those I think do). All the data should still be available anonymously,
it's just not all available anonymously through trac like it was before.
If you want to continue to browse files from git on the web through
trac, you can just create and verify an account to get access, or clone
the repo and browse locally, or browse anonymously on the web using one
of the other options listed at:
https://xapian.org/bleeding
I considered creating a guest account with all the viewing permissions
and advertising the username+passowrd on the front page but trac doesn't
seem to support locking down an account, so someone could log into this
account and mess with settings, change the password, or even delete it.
Let me know if you know of a better mitigation, or if there are
particular permissions that I've removed which are particularly useful
and don't result in high overhead.
Cheers,
Olly