[OpenAFS] Continuing problems with aklog on Mac OS X (rc10), internal state confused?
Mon, 03 Apr 2006 00:37:29 -0400
> To: firstname.lastname@example.org
> From: Adam Megacz <email@example.com>
> Organization: UC Berkeley
> Subject: [OpenAFS] Continuing problems with aklog on Mac OS X (rc10), internal state
> Date: Sun, 02 Apr 2006 20:27:35 -0700
> This is a real Heisenbug. From time to time, OpenAFS gets in this
> weird state where any aklog attempts give me this (kinit works fine,
> aklog: unable to obtain tokens for cell megacz.com (status: 11862790).
> But simply stopping and restarting the OpenAFS service fixes everything:
> megacz@maxwell:~$sudo /Library/StartupItems/OpenAFS/OpenAFS stop
> Stopping AFS
> Unmounting /afs
> Shutting down afsd processes
> Unloading AFS kernel extensions
> megacz@maxwell:~$sudo /Library/StartupItems/OpenAFS/OpenAFS start
> Starting OpenAFS
> Loading AFS kernel extensions
> kextload: /var/db/openafs/etc/afs.kext loaded successfully
> Starting afsd
> afsd: All AFS daemons started.
> megacz@maxwell:~$aklog -c megacz.com
> So, it looks like there's some internal state inside the OpenAFS
> kernel module (or afsd) that is getting confused. Anybody have any
> ideas on what that bit of state might be?
> - a
> PGP/GPG: 5C9F F366 C9CF 2145 E770 B1B8 EFB1 462D A146 C380
> OpenAFS-info mailing list
sisyphus% translate_et 11862790
11862790 (ktc).6 = unknown cell was passed to SetToken
I wouldn't care to venture why your cache manager is forgetting
all knowledge of your cell.
But I venture to guess that "fs newcell megacz.com <some-list-of-ip-addresses>"
might be a more precise fix than reloading the cache manager.