#3840 closed enhancement (fixed)
nfs-utils-1.2.7
Reported by: | Owned by: | ||
---|---|---|---|
Priority: | normal | Milestone: | |
Component: | BOOK | Version: | SVN |
Severity: | normal | Keywords: | |
Cc: |
Description ¶
Version increment.
I've got certain issues with this, and also with libtirpc-0.2.3 which is already in the book. First noticed when I built 7.3-rc1, but I forgot about them until tonight when I tried to boot another 7.3 build.
- /etc/netconfig appears to be required. Without it I can't mount nfs. From daemon.log:
Apr 1 23:31:14 jtm1 rpcbind: could not read /etc/netconfig Apr 1 23:31:14 jtm1 rpc.statd[2219]: Version 1.2.7 starting Apr 1 23:31:14 jtm1 sm-notify[2220]: Version 1.2.7 starting Apr 1 23:31:14 jtm1 rpc.statd[2219]: Failed to access local netconfig database: Netconfig database not found Apr 1 23:31:14 jtm1 rpc.statd[2219]: failed to create RPC listeners, exiting
If I add a mimimal /etc/netconfig (I will attach it), this all goes away and I'm back in business.
- rpcinfo (from rpcbind-0.2.0, which is the same version I was using with LFS-7.2) no longer accepts '-p hostname' : I had that in ~/.bashrc to mount certain directories if my server was exporting them. Now I have to omit the '-p'.
[sigh/] I'll pick this up later, for the moment just documenting it and allowing people to comment.
Change History (5)
by , 12 years ago
comment:1 by , 12 years ago
Owner: | changed from | to
---|---|
Status: | new → assigned |
comment:2 by , 12 years ago
This is really doing my head in :-(
First, I can't find what changed to require /etc/netconfig. The manpage netconfig.5 comes from rpcinfo, but it did in 0.2.2 as well as current rpcinfo-0.2.3. I was previously using nfs-utils-1.2.5, the code in that and 1.2.7 refers to /etc/netconfig in all the same places. I did find non-recent distro references where people have needed it on linux (apparently, it used to come from, or get packaged in, libtirpc on Mandriva in 2009 - packager dropped it because he thought it wasn't needed). Will add it (for nfs clients).
Second, there is a testsuite but it doesn't work [ and needs to be run as root, while statd is not working ] - it ought to detect that statd is already runnng, but instead it fails. Stepping through it (_with_ nfs working) the problem code is
start_statd() { rpcinfo -u 127.0.0.1 status 1 &> /dev/null if [ $? -eq 0 ]; then echo "***ERROR***: statd is already running and should " echo " be down when starting this test" return 1 fi $srcdir/../utils/statd/statd --no-notify }
Running that rpcinfo command without backgrounding it or discarding the output, I get:
#rpcinfo -u 127.0.0.1 status 1 rpcinfo: RPC: Unknown host program 100024 version 1 is not available
That's *far* too much to summarise in the XML, I'll just label it as "does not have a working test suite".
minimal /etc/netconfig, based on dragonfly BSD version.