EADDRNOTAVAIL syslog floods on reconnect

From Linux NFS

(Difference between revisions)
Jump to: navigation, search
(Created page with '== About == * Kernel version: 2.6.29-ge0724bf6e4, 2.6.29.1, 2.6.30-rc1-git3, 2.6.30-rc2-git7 * [https://bugzilla.kernel.org/show_bug.cgi?id=13034 bug 13034] * Reported by: Petr V…')
 
Line 9: Line 9:
* dmesg fills with: <pre>xs_tcp_setup_socket: connect returned unhandled error -99 </pre>
* dmesg fills with: <pre>xs_tcp_setup_socket: connect returned unhandled error -99 </pre>
* machine becomes unusable
* machine becomes unusable
 +
** if hardware watchdog is running, the machine may be rebooted
== Resolution ==
== Resolution ==

Latest revision as of 15:30, 25 October 2010

About

  • Kernel version: 2.6.29-ge0724bf6e4, 2.6.29.1, 2.6.30-rc1-git3, 2.6.30-rc2-git7
  • bug 13034
  • Reported by: Petr Vandrovec (April 7, 2009)
  • Fixed by: Trond Myklebust (April 21, 2009)

Symptoms

  • ls -l in directory with a link to a NFS-mounted directory
  • dmesg fills with:
    xs_tcp_setup_socket: connect returned unhandled error -99 
  • machine becomes unusable
    • if hardware watchdog is running, the machine may be rebooted

Resolution

commit f75e6745aa3084124ae1434fd7629853bdaf6798
Author: Trond Myklebust <Trond.Myklebust@netapp.com>
Date:   Tue Apr 21 17:18:20 2009 -0400

    SUNRPC: Fix the problem of EADDRNOTAVAIL syslog floods on reconnect
    
    See http://bugzilla.kernel.org/show_bug.cgi?id=13034
    
    If the port gets into a TIME_WAIT state, then we cannot reconnect without
    binding to a new port.
    
    Tested-by: Petr Vandrovec <petr@vandrovec.name>
    Tested-by: Jean Delvare <khali@linux-fr.org>
    Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
    Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Personal tools