Bug spam from non-malicious user's account

I don't know if this is a bug in lp as such. We do see people lose
the security of their account sometimes.

I suspect the user got a virus or other machine compromise that ended
up replying to all mail in their inbox with spam messages.

Martin: I filed this bug as per our spam-monitoring procedure. See our maintenance guide. Whatever the case, it does need to be looked into and incorporated into our institutional knowledge about spam attacks.

What spam-monitoring? The one written several years ago? This is not a bug in our code. Unless this bug is about detecting impersonations, thus bug can be closed.

On 14 October 2011 01:08, Curtis Hovey <email address hidden> wrote:
> What spam-monitoring? The one written several years ago? This is not a
> bug in our code. Unless this bug is about detecting impersonations, thus
> bug can be closed.

If, as I suspect, they lost control of their mail client, detecting
impersonation is going to be hard.

I had a brief look through the logs for whether these messages came
through mail, but I can't see a record either way. It's probably
there somewhere.

To prevent things like this we could screen new urls for spamminess,
but that doesn't seem like a critical problem.

By spam monitoring I mean the human eye on attack patterns, as documented on the spam-watching page from our maintenance-rotation docs.

I had a look at the database, and found a non-null Message.raw. According to our schema comments, that suggests that it came in by email.

I don't recall whether the policy to file a Critical bug for spam also extends to emailed comments, but evidently we have a need to figure out whether a comment came in through email or through the web UI.

Most comment spam comes from email. The sender is either an impersonator or has stolen the user credentials.

On 15 October 2011 00:25, Jeroen T. Vermeulen <email address hidden> wrote:
> I had a look at the database, and found a non-null Message.raw.
> According to our schema comments, that suggests that it came in by
> email.

perhaps for admins lp could identify mail messages in the ui in some
subtle way, like showing an envelope whose tooltip gives some data
about the message.

> I don't recall whether the policy to file a Critical bug for spam also
> extends to emailed comments, but evidently we have a need to figure out
> whether a comment came in through email or through the web UI.

I updated the policy to say you should file a bug if there is not one
already. Filing one per time spam is noticed, which is what it
literally said before, is obvious nonsense.