[OpenAFS-devel] salvager stuck on one file?

Miles Davis miles@CS.Stanford.EDU
Wed, 19 Apr 2006 21:10:40 -0700


Anybody ever see the salvager seemingly get stuck on one file? I've seen 
it twice today, same file -- it's a little under 2MB, and I can't tell 
what the hell it's doing. This is OpenAFS 1.4.0 on linux (fedora core 
4), reiserfs. Strace shows it chewing on the file:

llseek(7, 0, [0], SEEK_SET)            = 0
flock(7, LOCK_EX)                       = 0
_llseek(7, 824838, [824838], SEEK_SET)  = 0
read(7, "\2\0", 2)                      = 2
_llseek(7, 824838, [824838], SEEK_SET)  = 0
read(7, "\2\0", 2)                      = 2
_llseek(7, 824838, [824838], SEEK_SET)  = 0
write(7, "\1\0", 2)                     = 2
fsync(7)                                = 0
flock(7, LOCK_UN)                       = 0
_llseek(7, 0, [0], SEEK_SET)            = 0
flock(7, LOCK_EX)                       = 0
_llseek(7, 824842, [824842], SEEK_SET)  = 0
read(7, "\2\0", 2)                      = 2
_llseek(7, 824842, [824842], SEEK_SET)  = 0
read(7, "\2\0", 2)                      = 2
_llseek(7, 824842, [824842], SEEK_SET)  = 0
write(7, "\1\0", 2)                     = 2
fsync(7)                                = 0
flock(7, LOCK_UN)                       = 0
_llseek(7, 0, [0], SEEK_SET)            = 0
flock(7, LOCK_EX)                       = 0
_llseek(7, 824846, [824846], SEEK_SET)  = 0
read(7, "\2\0", 2)                      = 2
_llseek(7, 824846, [824846], SEEK_SET)  = 0
read(7, "\2\0", 2)                      = 2
_llseek(7, 824846, [824846], SEEK_SET)  = 0

(fd 7 is /vicepe/AFSIDat/x/xc2+U/special/zzzzPoXG++0, if that helps)

Upping the debug level on the file server via -TSTP doesn't seem to 
provide any information. After a couple of hours it will finish. 

-- 
// Miles Davis - miles@cs.stanford.edu - http://www.cs.stanford.edu/~miles
// Computer Science Department - Computer Facilities
// Stanford University