[Bug 252913] Enable option TCPHPTS and WITH_EXTRA_TCP_STACKS=1 in the kernel causes a massive mbufs leaks

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=252913

            Bug ID: 252913
           Summary: Enable option TCPHPTS and WITH_EXTRA_TCP_STACKS=1 in
                    the kernel causes a massive mbufs leaks
           Product: Base System
           Version: 12.2-STABLE
          Hardware: amd64
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: kern
          Assignee: bugs at FreeBSD.org
          Reporter: iron.udjin at gmail.com

OS: 12.2-STABLE stable/12-c243266-gd970a8218e16

In kernel conf:
makeoptions WITH_EXTRA_TCP_STACKS=1
options TCPHPTS

"netstat -m" in a few weeks of usage under high network load:

78365279/16/78365295 mbufs in use (current/cache/total)
43996/18104/62100/12244576 mbuf clusters in use (current/cache/total/max)
2429/16 mbuf+clusters out of packet secondary zone in use (current/cache)
12584/2975/15559/6122288 4k (page size) jumbo clusters in use
(current/cache/total/max)
0/0/0/1814011 9k jumbo clusters in use (current/cache/total/max)
0/0/0/1020381 16k jumbo clusters in use (current/cache/total/max)
19729647K/48112K/19777759K bytes allocated to network (current/cache/total)
42375955/0/0 requests for mbufs denied (mbufs/clusters/mbuf+clusters)
404032/0/33003 requests for mbufs delayed (mbufs/clusters/mbuf+clusters)
0/0/0 requests for jumbo clusters delayed (4k/9k/16k)
0/0/0 requests for jumbo clusters denied (4k/9k/16k)
1899771 sendfile syscalls
399 sendfile syscalls completed without I/O request
1004979 requests for I/O initiated by sendfile
7900311 pages read by sendfile as part of a request
399 pages were valid at time of a sendfile request
0 pages were valid and substituted to bogus page
0 pages were requested for read ahead by applications
0 pages were read ahead by sendfile
0 times sendfile encountered an already busy page
0 requests for sfbufs denied
0 requests for sfbufs delayed

As you can see, almost 20Gb of RAM allocated for network. Finally server
becomes unreachable: [zone: mbuf] kern.ipc.nmbufs limit reached

Additional info and counters:
https://forums.freebsd.org/threads/mbufs-leaks.78480/

-- 
You are receiving this mail because:
You are the assignee for the bug.