On 06/27/2013 05:34 PM, Ben Eisenbraun wrote:
Howdy Y'all,
Howdy.
It looks like REFMAC is trying to open
$CLIBD_MON/list/mon_lib_list.cif read-write and then read-only,
Yes, it does... curious. In libcheck.f's WRT_LIB_LIST_NEW_STYLE, should
be OPENFR(), I imagine.
but
with the flag to create it, which is causing the system to return -1
for both open calls.
stat("/programs/x86_64-linux/ccp4/6.3.0/ccp4-6.3.0/lib/data/monomers/list/mon_lib_list.cif",
{st_mode=S_IFREG|0644, st_size=1078257, ...}) = 0
fstat(2, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
fstat(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
fstat(0, {st_mode=S_IFREG|0600, st_size=750, ...}) = 0
open("/programs/x86_64-linux/ccp4/6.3.0/ccp4-6.3.0/lib/data/monomers/list/mon_lib_list.cif",
O_RDWR|O_CREAT, 0666) = -1 EROFS (Read-only file system)
open("/programs/x86_64-linux/ccp4/6.3.0/ccp4-6.3.0/lib/data/monomers/list/mon_lib_list.cif",
O_RDONLY|O_CREAT, 0666) = -1 EROFS (Read-only file system)
lseek(1, 0, SEEK_CUR) = -1 ESPIPE (Illegal seek)
lseek(1, 0, SEEK_CUR) = -1 ESPIPE (Illegal seek)
write(1, " Open failed: Unit: 7, File: /"..., 212) = 212
lseek(1, 0, SEEK_CUR) = -1 ESPIPE (Illegal seek)
lseek(1, 0, SEEK_CUR) = -1 ESPIPE (Illegal seek)
write(1, "<B><FONT COLOR=\"#FF0000\"><!--SUM"..., 46) = 46
dup(2) = 5
fcntl(5, F_GETFL) = 0x8002 (flags O_RDWR|O_LARGEFILE)
fstat(5, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,
0) = 0x2b4842405000
lseek(5, 0, SEEK_CUR) = -1 ESPIPE (Illegal seek)
write(5, "Last system error message: Illeg"..., 40) = 40
close(5) = 0
munmap(0x2b4842405000, 4096) = 0
write(2, " Refmac_5.7.0032: Open failed:"..., 147) = 147
What's confusing is that we have almost this exact same set up (RO NFS
sharing the software to many workstations) in _hundreds_ of labs and
none of them are having this problem.
I wonder if this is a (server-side?) SELinux issue.
I suppose I will go to the REFMAC/CCP4lib source next. Any other ideas
on how we can proceed?
restart the NFS? :-)
Other than that, fingers crossed for a one-letter fix (just tried it and
it didn't seem to damage anything (but then again the owner of the
process was the owner of mon_lib_list.cif...))
HTH,
Paul.
