Messages

[killall crowdsec]

Crowdsec running ? Try killall crowdsec. If not check if you're not out of space with df -h

# opnsense-update -zkr 25.1.3-fixlog had fixed it prior

It is expected that an update will revert to the latest released kernel - so when running a custom version with a particular patch you need to either confirm the new one has the fix included or get in the GUI and lock the kernel before the updates.

The good news here is that the patched kernel is still available and you can simply run the command and reboot

Disable OR fix IPv6 and it will work fine.

Great so Suricata is fine, this was just collateral damage. 25.1.4_1 was this patch

https://forum.opnsense.org/index.php?action=post;quote=233338;topic=46556

Does it work as expected if you revert it ?

Code Select Expand

opnsense-revert -r 25.1.3 suricata

Either you switch to OPNsense Business Edition or you keep up with the upgrade cycle on OPNsense CE.

OPNsense BE 25.4 will likely be based off of 25.1.4 which is supposed to arrive in a few days.

Holding off arbitrarily on older versions is not a secure option. If you have an actual issue report it here or on Github so it will be addressed one way or another, but don't expect things to magically start working whenever you feel like doing another upgrade. _Actual_ OPNsense bugs in core are extremely rare and far between if you care to look on Github, more often than not are "upstream" issues or misconfigurations that create the most noise. And those actual bugs are usually dealt with and a fix is provided in under 24h - from my observations, most often than not even if the report came in during a weekend.

If all goes well I'd expect a dnsmasq update as soon as next week.

https://www.freshports.org/dns/dnsmasq

Code Select Expand

version 2.91
    Fix spurious "resource limit exceeded messages". Thanks to
    Dominik Derigs for the bug report.

    Fix out-of-bounds heap read in order_qsort().
    We only need to order two server records on the ->serial field.
    Literal address records are smaller and don't have
    this field and don't need to be ordered on it.
    To actually provoke this bug seems to need the same server-literal
    to be repeated twice, e.g., --address=/a/1.1.1.1 --address-/a/1.1.1.1
    which is clearly rare in the wild, but if it did exist it could
    provoke a SIGSEGV. Thanks to Daniel Rhea for fuzzing this one.

    Fix buffer overflow when configured lease-change script name
    is too long.
    Thanks to Daniel Rhea for finding this one.

    Improve behaviour in the face of non-responsive upstream TCP DNS
    servers. Without shorter timeouts, clients are blocked for too long
    and fail with their own timeouts.

    Set --fast-dns-retries by default when doing DNSSEC. A single
    downstream query can trigger many upstream queries. On an
    unreliable network, there may not be enough downstream retries
    to ensure that all these queries complete.

    Improve behaviour in the face of truncated answers to queries
    for DNSSEC records. Getting these answers by TCP doesn't now
    involve a faked truncated answer to the downstream client to
    force it to move to TCP. This improves performance and robustness
    in the face of broken clients which can't fall back to TCP.

    No longer remove data from truncated upstream answers. If an
    upstream replies with a truncated answer, but the answer has some
    RRs included, return those RRs, rather than returning and
    empty answer.

    Fix handling of EDNS0 UDP packet sizes.
    When talking upstream we always add a pseudo header, and set the
        UDP packet size to --edns-packet-max. Answering queries from
    downstream, we get the answer (either from upstream or local
    data) If local data won't fit the advertised size (or 512 if
    there's not an EDNS0 header) return truncated. If upstream
        returns truncated, do likewise. If upstream is OK, but the
    answer is too big for downstream, truncate the answer.

    Modify the behaviour of --synth-domain for IPv6.
    When deriving a domain name from an IPv6 address, an address
    such as 1234:: would become 1234--.example.com, which is
    not legal in IDNA2008. Stop using the :: compression method,
    so 1234:: becomes
    1234-0000-0000-0000-0000-0000-0000-0000.example.com

    Fix broken dhcp-relay on *BSD. Thanks to Harold for finding
    this problem.

    Add --dhcp-option-pxe config. This acts almost exactly like
    --dhcp-option except that the defined option is only sent when
    replying to PXE clients. More importantly, these options are sent
    in reply PXE clients when dnsmasq in acting in PXE proxy mode. In
    PXE proxy mode, the set of options sent is defined by the PXE standard
    and the normal set of options is not sent. This config allows arbitrary
    options in PXE-proxy replies. A typical use-case is to send option
    175 to iPXE. Thanks to Jason Berry for finding the requirement for
    this.

    Support PXE proxy-DHCP and DHCP-relay at the same time.
        When using PXE proxy-DHCP, dnsmasq supplies PXE information to
        the client, which also talks to another "normal" DHCP server
        for address allocation and similar. The normal DHCP server may
        be on the local network, but it may also be remote, and accessed via
        a DHCP relay. This change allows dnsmasq to act as both a
        PXE proxy-DHCP server AND a DHCP relay for the same network.

    Fix erroneous "DNSSEC validated" state with non-DNSSEC
    upstream servers.  Thanks to Dominik Derigs for the bug report.

    Handle queries with EDNS client subnet fields better. If dnsmasq
    is configured to add an EDNS client subnet to a query, it is careful
    to suppress use of the cache, since a cached answer may not be valid
    for a query with a different client subnet. Extend this behaviour
    to queries which arrive a dnsmasq already carrying an EDNS client
    subnet.

    Handle DS queries to auth zones. When dnsmasq is configured to
    act as an authoritative server and has an authoritative zone
    configured, and receives a query for that zone _as_forwarder_
    it answers the query directly rather than forwarding it. This
    doesn't affect the answer, but it saves dnsmasq forwarding the
    query to the recursor upstream, which then bounces it back to dnsmasq
    in auth mode. The exception should be when the query is for the root
    of zone, for a DS RR. The answer to that has to come from the parent,
    via the recursor, and will typically be a proof-of-non-existence
    since dnsmasq doesn't support signed zones. This patch suppresses
    local answers and forces forwarding to the upstream recursor for such
    queries. It stops breakage when a DNSSEC validating client makes
    queries to dnsmasq acting as forwarder for a zone for which it is
    authoritative.

    Implement "DNS-0x20 encoding", for extra protection against
    reply-spoof attacks. Since DNS queries are case-insensitive,
    it's possible to randomly flip the case of letters in a query
    and still get the correct answer back.
    This adds an extra dimension for a cache-poisoning attacker
    to guess when sending replies in-the-blind since it's expected
    that the legitimate answer will have the same  pattern of upper
    and lower case as the query, so any replies which don't can be
    ignored as malicious. The amount of extra entropy clearly depends
    on the number of a-z and A-Z characters in the query, and this
    implementation puts a hard limit of 32 bits to make resource
    allocation easy. This about doubles entropy over the standard
    random ID and random port combination. This technique can interact
    badly with rare broken DNS servers which don't preserve the case
    of the query in their reply. The first time a reply is returned
    which matches the query in all respects except case, a warning
    will be logged. In this release, 0x020-encoding is default-off
    and must be explicitly enabled with --do-0x20-encoding. In future
    releases it may default on. You can avoid a future release
    changing the behaviour of an installation with --no-x20-encode.
   
    Fix a long-standing problem when two queries which are identical
    in every repect _except_ case, get combined by dnsmasq. If
    dnsmasq gets eg, two queries for example.com and Example.com
    in quick succession it will get the answer for example.com from
    upstream and send that answer to both requestors. This means that
    the query for Example.com will get an answer for example.com, and
    in the modern DNS, that answer may not be accepted.


Ah sorry, think I know now what that's for

I don't think */30 is a valid entry, or if it is then you got yourself a block from the list owner(s) for hammering their servers. Usually once a day is acceptable for most list providers.

Can you try this kernel as well ?

Code Select Expand

opnsense-update -zkr 25.1.3-states

Assuming everything is configured correctly, go in Rules - OpenVPN and send the traffic to the desired gateway, else by default it will leave on WAN

Did you install the plugin ? GUI or SSH would work just fine

Code Select Expand

pkg install os-realtek-re

OK, let's try this.

Set the mirror to Amsterdam. Reboot - in case there are stuck processes.

Then.

Code Select Expand

pkg install -f pkg && opnsense-update -bk
If it completes successfully and asks for a reboot do it, else post the output here.

When it's back.

Code Select Expand

opnsense-update -p
Same as in the previous step, if it completes successfully reboot, else post the output.

Do a healthcheck and post the full output here.

Corruption is highly unlikely.

If everything is working for you on 25.1 and stops on 25.1.3 it should be properly diagnosed and addressed, however you haven't provided any details abour your setup and waiting on 25.1 won't solve anything.

Having IPv6 off certainly clears up quite a few troubleshoothing scenarios, but other than that there are no reports of a remotely similar issue in 25.1.1-25.1.3 that I know of, so don't expect magical fixes as they're not coming.


Other than the description of your network setup that is the most important now, do you know if this issue was present in 25.1.1/25.1.2 ?