Skip to contentSkip to conversation

New: SCTP responds to heartbeat from the wrong address / interface

Bug 205439 Stephen Hemminger 3 participants 3 messages Nov 5, 2019 – Nov 7, 2019 SCTP protocol

lore
Share & tools
lore
#0 lore raw

Begin forwarded message:

Date: Tue, 05 Nov 2019 11:59:05 +0000
From: [email protected]
To: [email protected]
Subject: [Bug 205439] New: SCTP responds to heartbeat from the wrong address / interface

https://bugzilla.kernel.org/show_bug.cgi?id=205439

            Bug ID: 205439
           Summary: SCTP responds to heartbeat from the wrong address /
                    interface
           Product: Networking
           Version: 2.5
    Kernel Version: 4.9.35-server
          Hardware: Intel
                OS: Linux
              Tree: Mainline
            Status: NEW
          Severity: high
          Priority: P1
         Component: Other
          Assignee: [email protected]
          Reporter: [email protected]
        Regression: No

Seems that SCTP responds to heartbeat from the wrong address / interface.

# netstat -nS | fgrep 162.93.39.
sctp       0      0 81.226.10.66:3561     162.93.39.198:3561      ESTABLISHED
                    81.226.10.70          162.93.39.194
sctp       0      0 81.226.10.66:3560     162.93.39.134:3560      ESTABLISHED
                    81.226.10.70          162.93.39.130

Bad:

    3   0.055267 162.93.39.130 → 185.226.110.70 SCTP 74 HEARTBEAT
    4   0.055291 81.226.10.66 → 162.93.39.130 SCTP 74 HEARTBEAT_ACK
    7   0.355746 162.93.39.194 → 185.226.110.70 SCTP 74 HEARTBEAT
    8   0.355758 81.226.10.66 → 162.93.39.194 SCTP 74 HEARTBEAT_ACK

Good:

   13   0.551876 162.93.39.134 → 81.226.10.66 SCTP 74 HEARTBEAT
   14   0.551906 81.226.10.66 → 162.93.39.134 SCTP 74 HEARTBEAT_ACK
   19   0.853376 162.93.39.198 → 81.226.10.66 SCTP 74 HEARTBEAT
   20   0.853388 81.226.10.66 → 162.93.39.198 SCTP 74 HEARTBEAT_ACK

This is multi-homed scenario of SCTP.

Regards,

Signature · 2 lines
You are receiving this mail because:
You are the assignee for the bug.
#1 lore raw
Quoted 50 lines · Stephen Hemminger

Begin forwarded message:

Date: Tue, 05 Nov 2019 11:59:05 +0000
From: [email protected]
To: [email protected]
Subject: [Bug 205439] New: SCTP responds to heartbeat from the wrong address / interface

https://bugzilla.kernel.org/show_bug.cgi?id 5439

            Bug ID: 205439
           Summary: SCTP responds to heartbeat from the wrong address /
                    interface
           Product: Networking
           Version: 2.5
    Kernel Version: 4.9.35-server
          Hardware: Intel
                OS: Linux
              Tree: Mainline
            Status: NEW
          Severity: high
          Priority: P1
         Component: Other
          Assignee: [email protected]
          Reporter: [email protected]
        Regression: No

Seems that SCTP responds to heartbeat from the wrong address / interface.

# netstat -nS | fgrep 162.93.39.
sctp       0      0 81.226.10.66:3561     162.93.39.198:3561      ESTABLISHED
                    81.226.10.70          162.93.39.194
sctp       0      0 81.226.10.66:3560     162.93.39.134:3560      ESTABLISHED
                    81.226.10.70          162.93.39.130

Bad:

    3   0.055267 162.93.39.130 → 185.226.110.70 SCTP 74 HEARTBEAT
    4   0.055291 81.226.10.66 → 162.93.39.130 SCTP 74 HEARTBEAT_ACK
    7   0.355746 162.93.39.194 → 185.226.110.70 SCTP 74 HEARTBEAT
    8   0.355758 81.226.10.66 → 162.93.39.194 SCTP 74 HEARTBEAT_ACK

Good:

   13   0.551876 162.93.39.134 → 81.226.10.66 SCTP 74 HEARTBEAT
   14   0.551906 81.226.10.66 → 162.93.39.134 SCTP 74 HEARTBEAT_ACK
   19   0.853376 162.93.39.198 → 81.226.10.66 SCTP 74 HEARTBEAT
   20   0.853388 81.226.10.66 → 162.93.39.198 SCTP 74 HEARTBEAT_ACK
Quoted 4 lines

This is multi-homed scenario of SCTP.

Regards,

According to the RFC:
https://tools.ietf.org/html/rfc4960#page-38
In section 3.3.6, theres no guarantee that a HEARTBEAT_ACK will be sent from the same address that the HEARTBEAT chunk was received on, only that it will be sent back to the same address that generated the HEARTBEAT, which both cases above are doing.

If you want to guarantee source address selection in addition to the destination address guarantee, I think you need to set up your routing tables such that the source address to the given destination is specified.

We could probably change that to guarantee source address selection, but I'm not sure thats a universally desired feature.

Best
Neil

Quoted 4 lines
Signature · 2 lines
You are receiving this mail because:
You are the assignee for the bug.
#2 lore raw
In reply to Stephen Hemminger

Cc'ing bz reporter.

Quoted 78 lines · Neil Horman
Quoted 50 lines · Stephen Hemminger

Begin forwarded message:

Date: Tue, 05 Nov 2019 11:59:05 +0000
From: [email protected]
To: [email protected]
Subject: [Bug 205439] New: SCTP responds to heartbeat from the wrong address / interface

https://bugzilla.kernel.org/show_bug.cgi?id 5439

            Bug ID: 205439
           Summary: SCTP responds to heartbeat from the wrong address /
                    interface
           Product: Networking
           Version: 2.5
    Kernel Version: 4.9.35-server
          Hardware: Intel
                OS: Linux
              Tree: Mainline
            Status: NEW
          Severity: high
          Priority: P1
         Component: Other
          Assignee: [email protected]
          Reporter: [email protected]
        Regression: No

Seems that SCTP responds to heartbeat from the wrong address / interface.

# netstat -nS | fgrep 162.93.39.
sctp       0      0 81.226.10.66:3561     162.93.39.198:3561      ESTABLISHED
                    81.226.10.70          162.93.39.194
sctp       0      0 81.226.10.66:3560     162.93.39.134:3560      ESTABLISHED
                    81.226.10.70          162.93.39.130

Bad:

    3   0.055267 162.93.39.130 → 185.226.110.70 SCTP 74 HEARTBEAT
    4   0.055291 81.226.10.66 → 162.93.39.130 SCTP 74 HEARTBEAT_ACK
    7   0.355746 162.93.39.194 → 185.226.110.70 SCTP 74 HEARTBEAT
    8   0.355758 81.226.10.66 → 162.93.39.194 SCTP 74 HEARTBEAT_ACK

Good:

   13   0.551876 162.93.39.134 → 81.226.10.66 SCTP 74 HEARTBEAT
   14   0.551906 81.226.10.66 → 162.93.39.134 SCTP 74 HEARTBEAT_ACK
   19   0.853376 162.93.39.198 → 81.226.10.66 SCTP 74 HEARTBEAT
   20   0.853388 81.226.10.66 → 162.93.39.198 SCTP 74 HEARTBEAT_ACK
Quoted 4 lines

This is multi-homed scenario of SCTP.

Regards,

According to the RFC:
https://tools.ietf.org/html/rfc4960#page-38
In section 3.3.6, theres no guarantee that a HEARTBEAT_ACK will be sent from the same address that the HEARTBEAT chunk was received on, only that it will be sent back to the same address that generated the HEARTBEAT, which both cases above are doing.

If you want to guarantee source address selection in addition to the destination address guarantee, I think you need to set up your routing tables such that the source address to the given destination is specified.

We could probably change that to guarantee source address selection, but I'm not sure thats a universally desired feature.

Best
Neil

Quoted 4 lines
Signature · 2 lines
You are receiving this mail because:
You are the assignee for the bug.