[OpenAFS] Re: 1.6.2 buserver + butc

[Andrew Deason]

On Thu, 28 Mar 2013 08:33:26 -0400 (EDT)
Eric Sturdivant <sturdiva@umd.edu> wrote:

> -----------------  lwp# 8 / thread# 8  --------------------
>   ff17ac78 lwp_park (0, 0, 0)
>   ff174cc0 cond_wait_queue (46c338, 46c308, 0, 0, 1c00, 0) + 4c
>   ff175208 cond_wait (46c338, 46c308, 4, d7690, 46c2e0, 4) + 10
>   ff175244 pthread_cond_wait (46c338, 46c308, 0, 4, 5, 5) + 8
>   000989d8 rxi_ReadProc (46c2d0, fd84aa74, 4, d7690, 46c2e0, 4) + 640
>   0009c2ec ???????? (fd84ab38, fd84bb10, 0, fd84bb10, 46c2d0, 1)
>   00059b44 xdr_budb_dumpEntry (fd84ab38, fd84bb10, 1, 9b800, 9b800, 9b800) + 8
>   00053310 BUDB_CreateDump (447af8, fd84bb10, 0, fd84bb10, 46c2d0, 1) + 58
>   0005350c ubik_BUDB_CreateDump (150d, 46c230, fd84bb10, 0, 46c230, 0) + 1bc
>   000328dc createDump (2, 10f400, fd84b2b4, fd84bb10, fd84b6dc, 5) + e8
>   00030d94 Dumper   (4bfa28, 2f64d1, 0, 0, 4bfa38, 10f400) + 260
>   ff17abd8 _lwp_start (0, 0, 0, 0, 0, 0)

We're waiting for the buserver to respond to a BUDB_CreateDump. So,
presumably it is hanging on something. A quick glance at that suggests
to me that it's most likely hanging on waiting for a write lock to the
ubik database, which is presumably hanging because it's waiting for
another transaction to finish or something.

Maybe 'udebug' would be helpful, but maybe not. Otherwise, the other
stuff mentioned before (network capture, core, etc) is still probably
the way to go.

-- 
Andrew Deason
adeason@sinenomine.net