pmap_set failed trying to start fam w/ xinetd

kernel_bum · 03-01-2005, 06:26 PM

I'm having some problems starting FAM-2.6.8-12 with xinetd on Fedora Core 1. In began with courier-imap logging the following errors:

Code:

imapd: Check for proper operation and configuration
imapd: of the File Access Monitor daemon (famd).
imapd: Failed to create cache file: maildirwatch 
imapd: Error: Input/output error

The courier-imap FAQ stats that FAM needs to be running along with portmap. Portmap is running but when I try to restart xinetd I get this error:

Code:

pmap_set failed. service=sgi_fam program=391002 version=2

Here is the config for sgi_fam from xinted:

Code:

# default: on
# description: FAM is a file monitoring daemon. It can \
# be used to get reports when files change.
service sgi_fam
{
        type         = RPC UNLISTED
        socket_type  = stream
        user         = root
        group        = nobody
        server       = /usr/bin/fam
        wait         = yes
        protocol     = tcp
        rpc_version  = 2
        rpc_number   = 391002
        bind         = 127.0.0.1
}

pmap_sgi looks like it sets the list of registered RPC programs on the localhost. rpcinfo -p outputs:

Code:

program vers proto   port
 100000    2   tcp    111  portmapper
 100000    2   udp    111  portmapper
 391002    2   tcp    917  sgi_fam

Obviously, netstat -pnl | grep 917 shows there is nothing bound to that port. My understanding of rpc is limited and googling has either shown that running portmap solves the problem or I need to install FAM. Shutting down courier-imap and xinetd, then running fam -d -f outputs:

Code:

fam[19877]: log level is LOG_DEBUG
fam[19877]: read /etc/fam.conf line 13: "insecure_compatibility" = "false"
fam[19877]: read /etc/fam.conf line 20: "untrusted_user" = "nobody"
fam[19877]: read /etc/fam.conf line 28: "local_only" = "false"
fam[19877]: read /etc/fam.conf line 35: "xtab_verification" = "true"
fam[19877]: Setting untrusted-user to "nobody" (uid: 99, gid: 99)
fam[19877]: listening for clients on descriptor 3

Looks like it's working in standalone, but can't be started from xinetd. Any help is always appreciated.

sww · 10-13-2005, 12:05 AM

Had the same problem. Aparently the portmapper had an instance of it already registered. Here is how I resolved it.

pmap_dump > /tmp/x

edit the file /tmp/x and remove the sgi_fam line.

then stop portmap and restart THIS way

/etc/init.d/portmap stop
/etc/init.d/portmap start

reload the portmap

pmap_set < /tmp/x

then reactivate sgi_fam like this

chkconfig sgi_fam on

watch your /var/log/messages file and your should see no failures. If you do, then this was not your problem.

Sapient · 01-21-2006, 10:32 PM

I am attempting to install squirrelmail and aspell.

I tried the example above and here was my output:

file x:
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper

My logs are showing this:

Jan 21 20:24:13 mind imapd: Connection, ip=[::ffff:127.0.0.1]
Jan 21 20:24:13 mind imapd: LOGIN, user=sapient@hingedmind.com, ip=[::ffff:127.0.0.1], protocol=IMAP
Jan 21 20:24:13 mind imapd: Failed to create cache file: maildirwatch (sapient@hingedmind.com)
Jan 21 20:24:13 mind imapd: Error: Input/output error
Jan 21 20:24:13 mind imapd: Check for proper operation and configuration
Jan 21 20:24:13 mind imapd: of the File Access Monitor daemon (famd).
Jan 21 20:24:13 mind imapd: Failed to create cache file: maildirwatch (sapient@hingedmind.com)
Jan 21 20:24:13 mind imapd: Error: Input/output error
Jan 21 20:24:13 mind imapd: Check for proper operation and configuration
Jan 21 20:24:13 mind imapd: of the File Access Monitor daemon (famd).
Jan 21 20:24:13 mind imapd: LOGOUT, user=sapient@hingedmind.com, ip=[::ffff:127.0.0.1], headers=0, body=0, time=0