CentOS - May 2007 - CentOS 5 ssloooowww when network unavailable

This is undoubtedly an upstream issue, but ...

I've been messing about with that hp pavilion trying to get wireless
working.  If I boot with the eth0 NIC plugged in, everything is
blazingly fast -- GDM comes up in a few seconds, the Gnome startup
splash screen barely has time to appear before it's gone again, shells
pop up the instant I click Open Terminal.

If I unplug the network and reboot, it all goes to hell.  It takes
several minutes for GDM to start, the splash screen sometimes lingers
up to 10 minutes after the window manager is running, terminal windows
take varying lengths of time from 2-15 minutes to open.  It even takes
5-6 minutes to mount filesystems from the USB drive.

Nothing unusual appears in "top"; gnome-system-monitor is one of the
apps that takes forever to launch (to the point that I thought it had
crashed), but it doesn't show anything odd either.  No activity in the
various /var/log files.  Any ideas what could be causing this?

Sounds like DNS.

At a guess, I would try adding your hostname to the /etc/hosts file and 
set either a permanent IP address, or use 127.0.0.1

John.

Bart Schaefer wrote:
> This is undoubtedly an upstream issue, but ...
> 
> I've been messing about with that hp pavilion trying to get wireless
> working.  If I boot with the eth0 NIC plugged in, everything is
> blazingly fast -- GDM comes up in a few seconds, the Gnome startup
> splash screen barely has time to appear before it's gone again, shells
> pop up the instant I click Open Terminal.
> 
> If I unplug the network and reboot, it all goes to hell.  It takes
> several minutes for GDM to start, the splash screen sometimes lingers
> up to 10 minutes after the window manager is running, terminal windows
> take varying lengths of time from 2-15 minutes to open.  It even takes
> 5-6 minutes to mount filesystems from the USB drive.
> 
> Nothing unusual appears in "top"; gnome-system-monitor is one of
the
> apps that takes forever to launch (to the point that I thought it had
> crashed), but it doesn't show anything odd either.  No activity in the
> various /var/log files.  Any ideas what could be causing this?
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> http://lists.centos.org/mailman/listinfo/centos
> 
> 
> 

-- 
John Newbigin
Computer Systems Officer
Faculty of Information and Communication Technologies
Swinburne University of Technology
Melbourne, Australia
http://www.ict.swin.edu.au/staff/jnewbigin

Lorenzo wrote:
> John Newbigin ha scritto:
> > Sounds like DNS.
> > 
> > At a guess, I would try adding your hostname to the /etc/hosts file
> > and set either a permanent IP address, or use 127.0.0.1
> > 
> > John.
> > 
> I agree with John: the communications between the server and the
> client under X are trying to resolve the hostname which fails and
> they have to wait for the DNS query timeout.
> 
> The solution is to put the hostname on the 127.0.0.1 line in
> /etc/hosts If you want to be able to call the "official" address,
put
> the full hostname.domain on a separate line on /etc/hosts
> 
> I feel my english writing skills not adequate, so I write an example
> to explain better:
Don't apologize for it, your english writing is better than many people
I know here in the US.

-- 
Bowie

Bart,

I know it sounds strange, but.. ?Do you use the ntpd service? ?Is it active?

-- 
Thanks,
Jordi Espasa Clofent

I've been having a similar problem and wanted to provide what hopefully
is additional useful information.  Like the original poster, things seem
OK when booting with the network active.  However, network access is
extremely slow and timeouts are likely.  If I do a service network stop,
everything slows down, and keyboard and mouse activity is likely to be
dropped.  I also get error output which is included below.  Booting with
irqpoll doesn't seem to have any effect.

Of note is that my ethernet driver (r1000) is source built, not a driver
included with the build.  However, this driver was fine under CentOS 4.4.

Let me know what other information may be useful.  Thanks

Mark

-----
irq 177: nobody cared (try booting with the "irqpoll" option)
  [<c044aacb>] __report_bad_irq+0x2b/0x69
  [<c044acb8>] note_interrupt+0x1af/0x1e7
  [<c0570092>] usb_hcd_irq+0x23/0x50
  [<c044a2ff>] handle_IRQ_event+0x23/0x49
  [<c044a3d8>] __do_IRQ+0xb3/0xe8
  [<c04063f4>] do_IRQ+0x93/0xae
  [<c040492e>] common_interrupt+0x1a/0x20
  [<c045a83d>] __handle_mm_fault+0x7d/0x87b
  [<c04b8ea8>] avc_has_perm+0x3a/0x44
  [<c05fd4ef>] do_page_fault+0x20a/0x4b8
  [<c05fd2e5>] do_page_fault+0x0/0x4b8
  [<c0404a71>] error_code+0x39/0x40
  [<c04df804>] __copy_to_user_ll+0xd3/0xda
  [<c0429957>] do_proc_dointvec+0x1ee/0x291
  [<c0429a94>] proc_dointvec+0x15/0x19
  [<c0428c9c>] do_proc_dointvec_conv+0x0/0x36
  [<c0429134>] do_rw_proc+0xaa/0xee
  [<c042918e>] proc_readsys+0x0/0x13
  [<c042919e>] proc_readsys+0x10/0x13
  [<c046af3c>] vfs_read+0x9f/0x141
  [<c046b38a>] sys_read+0x3c/0x63
  [<c0403eff>] syscall_call+0x7/0xb
  ======================handlers:
[<c057006f>] (usb_hcd_irq+0x0/0x50)
Disabling IRQ #177