I tried the patch, but it didn't keep the fileserver from checking
for inodes and exiting.
afsconfig.h: #define AFS_NAMEI_ENV 1
param.sgi_65.h: #define AFS_SGI_XFS_IOPS_ENV 1 /* turns on
XFS inode ops. */
I paraphrase here, but after the patch:
#if !defined(AFS_LINUX20_ENV) && !defined(AFS_NT40_ENV)
...
#ifdef AFS_NAMEI_ENV
...
#else /* !AFS_NAMEI_ENV */
...
#endif
#ifdef AFS_SGI_XFS_IOPS_ENV
...
#endif
#endif
The define in param.sgi_65.h isn't a product of configure. So, the
XFS_IOPS section is being executed no matter what.
On Oct 11, 2005, at 7:33 PM, chas williams - CONTRACTOR wrote:
In message <431EFDB7-8068-458A-
[EMAIL PROTECTED]>,Jason Mader writes:
Tue Oct 11 17:30:13 2005 XFS/EFS File server starting
it probably should say "namei" instead of xfs/efs as well.
How come the /vicepa partition needs the larger inodes with the namei
fileserver?
it shouldnt. try this:
Index: src/vol/partition.c
===================================================================
RCS file: /cvs/openafs/src/vol/partition.c,v
retrieving revision 1.31
diff -u -u -r1.31 partition.c
--- src/vol/partition.c 19 Aug 2004 00:22:38 -0000 1.31
+++ src/vol/partition.c 11 Oct 2005 23:32:04 -0000
@@ -341,18 +341,18 @@
}
closedir(dirp);
}
-#else /* AFS_NAMEI_ENV */
+#else /* !AFS_NAMEI_ENV */
if (afs_stat(AFSIDatPath, &status) == 0) {
Log("This program is compiled without AFS_NAMEI_ENV, but
partition %s seems to contain volumes which use the namei-
interface; aborting\n", part);
return -1;
}
#endif /* AFS_NAMEI_ENV */
-#endif
#ifdef AFS_SGI_XFS_IOPS_ENV
if (VerifyXFSInodeSize(part, status.st_fstype) < 0)
return -1;
#endif
+#endif
#if defined(AFS_DUX40_ENV) && !defined(AFS_NAMEI_ENV)
if (status.st_ino != ROOTINO) {
_______________________________________________
OpenAFS-info mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-info
