I'm getting thousands of these a day, most times (99.99%) with internal sources, sometimes with an external source.Firmware is 17.1.3 MR3
We are getting thousands of these per day as well. I suspect it was affecting functionality on some of the sites our users visit. They were complaining of intermittent time-outs. Support was able to change IPS to "detect" versus "drop" somehow in the CLI even though IPS was diasabled on the rules in question. He seemed to realize quickly it was a known issue and escalated my case after grabbing some logs. v17.1.3 MR-3
Any chance you can post the rule responsible for this? I can't seem to find it.
Hi Ryan Collis
[Update] This KBA has been published for this issue.
That was part of the issue that was frustrating. Even if an IPS policy wasn't applied to the firewall rule in question it would still interfere with traffic and the user would experience time-outs on business critical websites, etc. So basically it's buggy IPS in itself generating 250,000+ "reset outside window" events AND the bugged IPS it's being applied when it shouldn't be.
Support was able to change IPS to "alert" versus "drop" in the CLI and that got users going again but they've since escalated the case to the "global escalation specialists". Obviously we want IPS working as intended so it's only a workaround.
Some other potential false positives with IPS we see with possible effects on legit traffic:"Data sent on stream after TCP Reset received" = 40,000+ times/day"TCP Timestamp is missing" = 10,000+ times/day"...Lets Encrypt SSL cert.." = 5000+ times/dayWe have about 250 users at this location.
Will update as we find more.
I am getting these as well, I did follow the KB, what confusing me is I am on XG105w 17.5.3 MR-3. So I would think Sophos would have fixed this by now without me having to use the KB. Since they still have the KB in place and are not sending the fix out with the latest release I am assuming that this is not completely fixed and the KB is a workaround until Sophos can figure out why there are so many false postives?
What I am looking for here is why was it enabled in the first place and what threats am I exposing the network too if I disable it?
The fix to this issue was the setting being disabled by default starting with SFOS v17.1.4 MR-4, so I apologize as it seems this did not occur for you.
As I mentioned previously:
This specific IPS signature has been disabled by default, starting with SFOS v17.1.4 MR-4 due to customers experiencing excessive false-positives.
These IPS signatures are triggered by TCP anomalies (includes RST packets received outside of window). This was causing some customers to experience valid RST packets being false-positively dropped.
Customers still experiencing excessive false-positives should raise a support case for further investigation. However, this setting can also be disabled via the console command (set ips tcp_option detect_anomalies disable) to allow the TCP anomaly decision to be made by the host client OS instead if desired.
Please continue to monitor and let me know if you run into any further issues.
It is not true that TCP anomalies is disabled starting with SFOS v17.1.4 MR-4, at least on my Sophos Home machine.
I fresh installed it using 17.5.6 MR-5 ISO without loading any previous backup. It is automatically updated to MR6 when I active the license.
Login to console and found that "var DETECT_ANOMALIES" is set to "yes"
console> show ips_confconfig stream 1config maxsesbytes 0config stdsig 1config qnum 10config maxpkts 8config disable_tcpopt_experimental_drops 0config enable_appsignatures 1var SEARCH_METHOD ac-qvar SIP_STATUS enabledvar IGNORE_CALL_CHANNEL enabledvar TCP_POLICY windowsvar LOCAL_RULE local.rulesvar DETECT_ANOMALIES yesvar TCP_BLOCK blockconfig failclose offconfig cpulist 0:1
Run suggested command to disable anomalies detection:
console> set ips tcp_option detect_anomalies disable
Check the ips_conf again var DETECT_ANOMALIES" is set to "no"
console> show ips_confconfig stream 1config maxsesbytes 0config stdsig 1config qnum 10config maxpkts 8config disable_tcpopt_experimental_drops 0config enable_appsignatures 1var SEARCH_METHOD ac-qvar SIP_STATUS enabledvar IGNORE_CALL_CHANNEL enabledvar TCP_POLICY windowsvar LOCAL_RULE local.rulesvar TCP_BLOCK nblockconfig failclose offconfig cpulist 0:1var DETECT_ANOMALIES no
The bug still didn't fix in V17.5.6 MR-6.User need to fix it with the console command...