Bug 56010

Summary: wrong order of init scripts
Product: [Retired] Red Hat Linux Reporter: Dr. Tilmann Bubeck <tilmann>
Component: kernel-pcmcia-csAssignee: Dave Jones <davej>
Status: CLOSED WONTFIX QA Contact: Brock Organ <borgan>
Severity: medium Docs Contact:
Priority: medium    
Version: 7.2CC: pfrields
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2004-11-25 08:14:41 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Dr. Tilmann Bubeck 2001-11-10 15:31:23 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i586; de-DE; rv:0.9.2)
Gecko/20010726 Netscape6/6.1

Description of problem:
the scripts in /etc/rc.d/rc#.d start the program as follows:

K03rhnsd   K74ntpd      S10network   S25netfs       S60lpd       S95atd
K20nfs     K96irda      S12syslog    S26apmd        S80sendmail  S99local
K20rwhod   S05kudzu     S13portmap   S28autofs      S85gpm
K46radvd   S08ipchains  S14nfslock   S45pcmcia      S90crond
K50snmpd   S08iptables  S17keytable  S56rawdevices  S90xfs
K65identd  S09isdn      S20random    S56xinetd      S95anacron

This means, that NIC in pcmcia will be started as #45.

The problem is, that many script before 45 need a working network:
  S12syslog (@ notation)
  S25netfs  (NFS filesystems)
  S28autofs (maps from NIS or LDAP)
  .. maybe more?

So I would suggest, that you move S45pcmcia to S11pcmcia, right _after_
S10network. So you still "delay" eth initialisation in case of pcmcia, but
the network will be running for all subsequent services.





Version-Release number of selected component (if applicable):


How reproducible:
Always

Steps to Reproduce:
1. Boot with pcmcia NIC
2. use autofs with LDAP or NIS maps

 

Actual Results:  autofs says "No mountpoints defined"

Expected Results:  autofs should use mount points defined in LDAP (which it
cannot, because the network is not yet activated).

Additional info:

This bug has caused many similar bug reports in bugzilla where users of
several other components (e.g. autofs) complain about various problem with
their component. These are often results of the late network initialisation
with pcmcia.

Comment 1 Need Real Name 2002-10-31 21:22:04 UTC
This is really not a bug.  Nothing in the boot sequence should depend on PCMCIA 
devices being configured first, as there is no reason to expect a PCMCIA device 
to even be present at boot time.

Moving the script doesn't even solve the problem.  Because cards are only 
powered up once the PCMCIA initialization script runs, it may take several 
seconds for cards to be initialized and identified, the correct drivers loaded, 
network interfaces to autonegotiate link speed, etc.  So running other scripts 
that assume a working network will give unpredictable results.

-- Dave Hinds


Comment 2 Dr. Tilmann Bubeck 2002-11-01 11:30:09 UTC
First of all, this bug report is quite old and corresponds to RH 7.2. At least
in 8.0 they fixed the thing by moving PCMCIA to #24 which solves all the
mentioned problems (except "@" in syslog).

But Dave is also right, the cardmgr, and drivers need some time, so that is is
not guaranteed, that you get a working pcmcia network when ending
/etc/init.d/pcmcia. I solved this problem by inserting a "sleep" at the end of
/etc/init.d/pcmcia to give the system some time to initialize.

Dave, you say, that nothing in the boot sequence should depend on pcmcia. Well,
how do you suggest to initialize a system with a NIC in pcmcia? What would you
change?

Till