Topics

moderated Virus scanning


Chris Jones
 

On Thu, Feb 21, 2019 at 06:59 PM, Mark Fletcher wrote:
 

Does your virus scanning detect problems in a zipped attachment?

Yes it does.
Mark; thanks for the prompt reply! :)

Chris


Sandra Pickens-Gmail
 

Mark,

I was the group owner that approved an email from a member that had a zip file attached.  I do not know if the zip file was a virus because I did not open the zip file. The group is FriedbergFriends.  Anything you need from me to investigate?  I removed the email from the group, but still have a copy in my personal email.

 

Sandra Pickens

 

From: main@beta.groups.io <main@beta.groups.io> On Behalf Of Mark Fletcher
Sent: Thursday, February 21, 2019 1:58 PM
To: main@beta.groups.io
Subject: Re: [beta] Virus scanning

 

On Thu, Feb 21, 2019 at 10:06 AM Chris Jones via Groups.Io <chrisjones12=btinternet.com@groups.io> wrote:


Does your virus scanning detect problems in a zipped attachment?

Yes it does.

 

I would like to know if a virus-infected file did get through the virus scanning. Someone recently contacted support about one such incident, but I haven't been able to verify it yet. From the logs, we don't actually see many virus infected files.

 

 

Thanks,
Mark

 


 

On Thu, Feb 21, 2019 at 10:06 AM Chris Jones via Groups.Io <chrisjones12=btinternet.com@groups.io> wrote:

Does your virus scanning detect problems in a zipped attachment?

Yes it does.

I would like to know if a virus-infected file did get through the virus scanning. Someone recently contacted support about one such incident, but I haven't been able to verify it yet. From the logs, we don't actually see many virus infected files.


Thanks,
Mark
 


Chris Jones
 

On Tue, Aug 21, 2018 at 04:25 AM, Mark Fletcher wrote:
My default implementation would be to turn it on so that it blocks all emails, files and photos that it finds has a virus or phishing attempt.
Mark; this thread has been referenced on GMF following someone reporting that a group member (actual group not reported) had had their email spoofed and a zipped attachment sent to the group.

Does your virus scanning detect problems in a zipped attachment?

Chris


 

HI All,

I've just enabled the new virus scanning option for groups. Under Message Policies, there's a dropdown labeled Viruses. It defaults to Blocked, but can be set to Moderated.

I disabled scanning for phishing schemes, because the false positive rate was just way too high. So it only scans for viruses. Emails from non-members that contain viruses are blocked regardless of the group setting. When a message is blocked or moderated, an entry is added to the activity log (there are 3 new actions: non-member blocked, member blocked, member moderated). When viewing a moderated message for approval, a red badge is displayed with the name of the virus detected. I'm not currently retroactively scanning the archives and I'm not currently displaying anything when viewing moderated and then approved messages in the archives.

In my testing, very few viruses are actually sent through the system. So, hopefully, this is something you won't encounter.

Please let me know if you have any questions.

Thanks,
Mark


 

Mark,

Honestly not sure what, if anything, to do with SPF and DKIM. Seeing
plenty of valid emails with bad DKIM sigs, for example.
I have two goals:

1) Prevent abuse posted to a group such as that I reported to support 8/11. This is where a malcontent or crook uses credentials at an otherwise legit mailbox provider to spoof a group member's address, resulting in junk posted to the group.

2) Eliminate the need for a confirmation email and response for email commands in most cases.

Case (1) is the more important, but I think (2) is easier. Mostly because there is no adverse consequence of a failed authentication in case (2) - you just send the confirmation email as you do now. You could accept the OR of two tests: a valid DKIM, signed by the header-from domain as one test; or a passing SPF but only if the the header-from domain is aligned with the envelope-from domain as the other test. This happens to be the same as the DMARC criteria, I think. So in the case of a "pass" you send a notification of the command's acceptance and effect rather than a request for confirmation.

In case (1) the difficulty arises if both tests fail. That's sure to be true in the case I want to weed out, but I don't know how many legit messages might be affected. I think the case you cited to me on 8/13 would be one such; unless we (you) can figure out something he could tell you by way of account information that would let you make a third test that would pass his messages.

My inclination is to say that it is "good enough" if you force messages that fail both tests into the pending queue, as if sent by a non-subscriber (but with a different marking, of course). That would have allowed the group mod in my support case to discard the abuse before it hit the group.

Yes, that would be a pain for the second fellow, and his group mods. But he's already committed to switching Thunderbird to use the correct SMTP server, so that should solve his case. One down...

Shal


Walter Underwood
 

On Aug 23, 2018, at 10:27 AM, Mark Fletcher <markf@corp.groups.io> wrote:

Honestly not sure what, if anything, to do with SPF and DKIM. Seeing plenty of valid emails with bad DKIM sigs, for example.

It might be worth talking to the folk at FastMail. They are pretty active in the mail security standards. I use them for my mail.


wunder
Walter Underwood
wunder@...
http://observer.wunderwood.org/  (my blog)


 

On Wed, Aug 22, 2018 at 11:08 PM, Shal Farley <shals2nd@...> wrote:

> I haven't done anything with SPF and DKIM data yet.

One step at a time. Though I might have expected these before content scanning. But I may have a skewed view of their relative difficulty and effectiveness (eg: these don't apply to uploads).

Honestly not sure what, if anything, to do with SPF and DKIM. Seeing plenty of valid emails with bad DKIM sigs, for example.


Mark 


Jim Higgins
 

Received from Mark Fletcher at 8/23/2018 05:09 AM UTC:

Ok, so new group option for dealing with spam: either moderate or reject, with reject being the default. Rejected messages will be logged in the activity log. If I reject a message, should it bounce back to the sender, or should I blackhole it?

Don't bounce! REJECT during the SMTP transaction if possible. And if not possible, then just blackhole it.


If a message in the archives is flagged as having a virus or phishing attack, should I put a banner on the page saying so? (and should I go back through the archives doing scans)?

For viruses I'd prefer deletion. Given a settable option I'd choose deletion and take the tiny chance it's a false positive rather then set myself up to second guess the scanner. Scans for phishing based on keywords in message bodies are less reliable so a banner might be the thing for that.

Yes on scanning existing files/images (binaries) for viruses. Not sure scanning archived message text would provide much added benefit, but if you have the CPU horsepower, it can't hurt.


The fact that many/most groups don't accept messages from non-subscribers acts as a natural prevention for a lot of this crap.

That and also some groups don't accept attachments... and are plain text only. I've NEVER seen spam or viruses - or even phishing attempts - in plain text email with no attachments.


I don't accept smtp connections from IP addresses that don't have reverse DNS records. I use a few blocklists as well, for all connections to the site, not just email. I haven't done anything with SPF and DKIM data yet.

This is very good to know.

Jim H


Gerald Boutin <groupsio@...>
 

On Thu, Aug 23, 2018 at 11:19 AM, JohnF wrote:
I would lean toward dropping rather than bouncing spam/malware/phishing, just because the system on the other end might misclassify groups.io as the source of the malware.

If you do bounce it for malware, whatever triggered the malware alert (attachment or link to dangerous site) should be stripped from the bounce message to avoid this.

JohnF

I am 100% in favor of reject and not bouncing.

If you want more info - read on here: http://www.dontbouncespam.org/

--
Gerald


 

I would lean toward dropping rather than bouncing spam/malware/phishing, just because the system on the other end might misclassify groups.io as the source of the malware.

If you do bounce it for malware, whatever triggered the malware alert (attachment or link to dangerous site) should be stripped from the bounce message to avoid this.

JohnF


Sharon Villines
 

Rejecting is annoying, but aggressively scrubbing malware from multipliers like groups.io is good policy in my book. Moderators should hesitate and consider carefully a decision to take the "moderate" option. Letting a malevolent email loose on your group could destroy it. Look at the annoyance of automatic rejection as a price a small price paid for the convenience and pleasure you get from membership in a groups.io group.
How likely is it that a clean message will be identified as infected? We aren’t talking about spam which is identified on the basis of words used, etc. Right?

If the messages are being scanned for malware, is it safe to allow attachments? I have a large public list on which we have rejected attachments for years. Since my personal malware software finds viruses in my non-list email regularly, it certainly is still getting through whatever other email systems see it.

Sharon
----
Sharon Villines
TakomaDC@Groups.io
"Neighbors Talking to Neighbors”
Takoma Park DC and MD


 

Hi Mark,

If I reject a message, should it bounce back to the sender,
or should I blackhole it?
Reject (5xx) during inbound SMTP-session.

If a message in the archives is flagged as having a virus or phishing
attack, should I put a banner on the page saying so? (and should I go back
through the archives doing scans)?
At least give moderators the option to silence false positives which are inevitable, especially about phishing.

--
Lena


 

Dave,

Will there be an option to disable spam checking?
Keep in mind this isn't about simple spam. This is about content which may be malware-infected or a phishing attempt. Detecting these specific things is likely to be much less prone to false positives than a generic spam filter.

Shal


 

Mark,

If I reject a message, should it bounce back to the sender, or should
I blackhole it?
Given that the sender passed a reverse-DNS I'd say it is safe to bounce it back. Chances are the message was sent by a compromised account at an otherwise legit service. (And maybe the rejection lets them know they've got a problem user?).

If a message in the archives is flagged as having a virus or phishing
attack, should I put a banner on the page saying so?
On the whole I'd say "yes".

The counter-argument is that if the group's mods accepted the message then they might not appreciate having the (presumed false-positive) marking on the message. But I think that the members deserve to know that there was at least some doubt about this content.

(and should I go back through the archives doing scans)?
Optional, but probably a good idea. The question is what you'd do besides mark them. I think adding entries for them to the Activity log may be sufficient (for mods to go find them if they want).

I haven't done anything with SPF and DKIM data yet.
One step at a time. Though I might have expected these before content scanning. But I may have a skewed view of their relative difficulty and effectiveness (eg: these don't apply to uploads).

Shal


Dave Sergeant
 

I take it this has not been implemented yet Mark, I see no options for
dealing with spam other than the existing ones under 'spam control' for
which in our case 'restricted membership' is the only relevant one.

Will there be an option to disable spam checking? I strongly feel there
should due to false positives, especially in private groups which
rarely have spam issues. I only look in the activity log rarely and
only if I have issues, 'moderation' of posts is an unwelcome extra
burden which I would really not want to do. Our groups just don't
suffer from spam, I would not really want false positives to get in the
way of an otherwise excellently running service.

Dave

On 22 Aug 2018 at 22:09, Mark Fletcher wrote:

Ok, so new group option for dealing with spam: either moderate or
reject, with reject being the default. Rejected messages will be logged
in the activity log. If I reject a message, should it bounce back to the
sender, or should I blackhole it?

http://davesergeant.com


 

On Tue, Aug 21, 2018 at 9:42 PM, Shal Farley <shals2nd@...> wrote:

I don't think I would go as far as Lena suggests, and moderate them without that being a group option; I'm concerned that few group moderators would have the knowledge to make a safe decision for their group. A choice between "moderate" or "reject" might be useful, with "reject" the default.


Ok, so new group option for dealing with spam: either moderate or reject, with reject being the default. Rejected messages will be logged in the activity log. If I reject a message, should it bounce back to the sender, or should I blackhole it?

If a message in the archives is flagged as having a virus or phishing attack, should I put a banner on the page saying so? (and should I go back through the archives doing scans)?

 
By the way, I assume none of the above applies to the boatloads of absolute junk from invalid sources (malware-infected PCs and the like) that I presume you've been dropping all along. Those deserve the black hole treatment.

The fact that many/most groups don't accept messages from non-subscribers acts as a natural prevention for a lot of this crap. I don't accept smtp connections from IP addresses that don't have reverse DNS records. I use a few blocklists as well, for all connections to the site, not just email. I haven't done anything with SPF and DKIM data yet.

Thanks,
Mark

 


Marv Waschke
 

On Tue, Aug 21, 2018 at 09:42 PM, Shal Farley wrote:
I don't think I would go as far as Lena suggests, and moderate them without that being a group option; I'm concerned that few group moderators would have the knowledge to make a safe decision for their group. A choice between "moderate" or "reject" might be useful, with "reject" the default.
Agree with Shal on this. An email reflector like groups.io, is a multiplier-- it turns a single email into many emails. Cybercriminals love this. Send out a malevolent email to a single address and, depending on the size of the group, the poison goes to hundreds or thousands of potential victims. And those victims are predisposed to swallow the poison because it comes from a familiar source that they intentionally subscribed to. Even better if the email spoofs the name of a prominent member. This is a hacker's dream setup. These criminals are not nice people. Give them a chance and they will hurt you.

Rejecting is annoying, but aggressively scrubbing malware from multipliers like groups.io is good policy in my book. Moderators should hesitate and  consider carefully a decision to take the "moderate" option. Letting a malevolent email loose on your group could destroy it. Look at the annoyance of automatic rejection as a price a small price paid for the convenience and pleasure you get from membership in a groups.io group.

Best, Marv


 

Mark,

> My default implementation would be to turn it on so that it blocks all
> emails, files and photos that it finds has a virus or phishing
> attempt.
> Do you see any reason to not do it this way?

"Block" = "Drop" or "Reject"?

Drop is a very severe action, and I'm not entirely sure it should be done even with non-subscribers. With subscribers at the least I'd recommend "reject" (and add to the Activity log).

As Jim and some others, I was thinking maybe have a group option to put those from subscribers in the pending queue, prominently marked as containing potentially harmful content. This would serve the small fraction of groups who might be studying such things, or might be sharing harmless executable files that trigger a false positive.

I'm sympathetic because once long ago forwarding a message to abuse@... or spam@... was a common way certain senders requested that non-users (of their service) should report "bad" messages coming from their service. But I had an ISP that blocked suspicious messages outbound by me, so I couldn't send the requested report.

I don't think I would go as far as Lena suggests, and moderate them without that being a group option; I'm concerned that few group moderators would have the knowledge to make a safe decision for their group. A choice between "moderate" or "reject" might be useful, with "reject" the default.

By the way, I assume none of the above applies to the boatloads of absolute junk from invalid sources (malware-infected PCs and the like) that I presume you've been dropping all along. Those deserve the black hole treatment.

Shal


Jim Higgins
 

I don't expect scanning to change anything here for me since my groups use plain text and no attachments in the first place.

Actions? Perhaps DROP ALL and DROP ATTACHMENT ONLY should be available... as well as maybe some sort of "Quarantine for Review/Moderation by Group Owner" option.

Notification? Notify Owner, Notify Sender, Notify Both seem like decent options. Notify Owner (or both) seems to go hand in hand with an action of "Quarantine for Review..." (above).

Jim H



Received from Mark Fletcher at 8/21/2018 03:24 AM UTC:

Hi All,

I've been testing virus/phishing scanning the last few weeks and I'm pretty confident that it's catching what it should. In testing, it's scanning all emails, all uploaded files and photos. And right now, if the sender is not a subscriber, it drops any emails it finds has a virus or phishing attack.

My default implementation would be to turn it on so that it blocks all emails, files and photos that it finds has a virus or phishing attempt. Do you see any reason to not do it this way?

The scanner I'm using is here: <http://www.clamav.net/>http://www.clamav.net/

Thanks, Mark