I'm doing my testing locally, therefore I use 127.0.0.1,
but I believe this will happen with any numeric IP:
# cat /etc/exports
/ 127.0.0.1(rw,no_root_squash,async)
# rm /var/lib/nfs/etab
# exportfs -r
# cat /var/lib/nfs/etab
/ localhost(rw,async,wdelay,hide,nocrossmnt,secure,no_root_squash,no_all_squash,subtree_check,secure_locks,mapping=identity,anonuid=-2,anongid=-2)
Looks like bug to me, because I said explicitly that I allow 127.0.0.1 to mount /,
not it's hostname.
One of consequences of this bug is mount failures if mountd gets restarted while
name resolution isn't working (DNS server down etc).
In case this is indeed a bug, here is where it happens.
(I'm not sure I will be able to cook up a sane patch
so at least I'll try to explain how 'exportfs -r' did it):
int main(...)
-> int export_read(char *fname)
-> nfs_export* export_lookup(char *hname, char *path, int canonical)
-> nfs_client* client_lookup(char *hname, int canonical)
nfs-utils-1.0.7/support/export/client.c:
nfs_client *
client_lookup(char *hname, int canonical)
{
nfs_client *clp = NULL;
int htype;
struct hostent *hp = NULL;
htype = client_gettype(hname);
if (htype == MCL_FQDN && !canonical) {
struct hostent *hp2;
hp = gethostbyname(hname);
if (hp == NULL || hp->h_addrtype != AF_INET) {
xlog(L_ERROR, "%s has non-inet addr", hname);
return NULL;
}
/* make sure we have canonical name */
hp2 = hostent_dup(hp);
HERE ====> hp = gethostbyaddr(hp2->h_addr, hp2->h_length,
hp2->h_addrtype);
if (hp) {
Hmm... maybe we need to introduce MCL_NUMERIC_IP or something like that?
--
vda
-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs