Mark Adams wrote:

> I have a headless server catching backups in my closet. It's connected
> to my network wirelessly and generally does a fine job. Problem is,
> after some updates, it seems the machine will lose connectivity.
>
> Now, it's kind of hard to know what's going on because the machine is
> headless and I'm SSH'ing into it from this desktop. My first clue that
> something is wrong comes when I check logs for daily backups. They will
> report I/O errors or just show up blank depending on the log.




> Once I see that, I try to get a directory listing for mounted shares
> from this machine or just "fs" to see if I can access it. I invariably
> get I/O errors. The next step is to walk back to the closet and use the
> power button to bounce the machine.

At this point have you lost your SSH connection? if not why not restart the
machine through SSH rather than going and hitting the power button?



> I've checked the logs and from all I can gather, it looks as if the
> machine is hanging just after SAMBA initializes. Beyond that I have no
> idea what's happening.

Stop Samba from starting at boot and possibly other networking services but
leave the network itself and SSH to start, see if it reboots cleanly.
start services manually via ssh and check they all work and add them back
until you find out which one causes the errors.

Andrew Ed

Andrew Ed wrote:
> Mark Adams wrote:
>
>> I have a headless server catching backups in my closet. It's connected
>> to my network wirelessly and generally does a fine job. Problem is,
>> after some updates, it seems the machine will lose connectivity.
>>
>> Now, it's kind of hard to know what's going on because the machine is
>> headless and I'm SSH'ing into it from this desktop. My first clue that
>> something is wrong comes when I check logs for daily backups. They will
>> report I/O errors or just show up blank depending on the log.
>
>
>
>
>> Once I see that, I try to get a directory listing for mounted shares
>> from this machine or just "fs" to see if I can access it. I invariably
>> get I/O errors. The next step is to walk back to the closet and use the
>> power button to bounce the machine.
>
> At this point have you lost your SSH connection? if not why not restart the
> machine through SSH rather than going and hitting the power button?

Right, I get the I/O errors because connectivity is gone. I can't get
into the box with anything.

>
>
>
>> I've checked the logs and from all I can gather, it looks as if the
>> machine is hanging just after SAMBA initializes. Beyond that I have no
>> idea what's happening.
>
> Stop Samba from starting at boot and possibly other networking services but
> leave the network itself and SSH to start, see if it reboots cleanly.
> start services manually via ssh and check they all work and add them back
> until you find out which one causes the errors.

That's an idea, though there seems to be a chance that it's whatever
comes after samba that's hosing the system. I'm tied up today, but when
I get the chance I'll pursue that.

Thanks Andrew.

--
Mark E. Adams, 2004 -- drop the "dot" to email me.
http://adamslan.shyper.com -*- Mandriva User# 263042

CONSIDER: ===========---------,,,,,,,,,............. . . . . .
Conquest is easy. Control is not.
-- Kirk, "Mirror, Mirror", stardate unknown

--
Posted via a free Usenet account from http://www.teranews.com

Mark Adams wrote:
> I have a headless server catching backups in my closet. It's connected
> to my network wirelessly and generally does a fine job. Problem is,
> after some updates, it seems the machine will lose connectivity.
>
> Now, it's kind of hard to know what's going on because the machine is
> headless and I'm SSH'ing into it from this desktop.

Does it have a serial port? Do you own a serial cable?
If so you can get in with a terminal emulator.

> Running nmap against it shows something like this:

So something or other is hanging after the network starts and
ssh never comes up. Get the serial port working, then reboot to
failsafe, and manually run through the /etc/rc.d/rc2.d steps until
you hit the one that is waiting for keyboard input.

For what it's worth, I've had unending problems with Mandriva
deciding it wants to write over my configuration with some default
or other, which breaks the computer lab PCs every time. A few
days ago it was once again an uncalled for rewriting of
/etc/shorewall/rules, removing my rules, and putting in a single
include of rules.drakx. Sigh. At least I know have my findchanges.sh
script in place (see previous post) and can now find these
problems very quickly. Anyway, now all my rules are in
rules.drakx, but no doubt IT will eventually be rewritten by
something else :-( .

Regards,

David Mathog

>
> That's an idea, though there seems to be a chance that it's whatever comes
> after samba that's hosing the system. I'm tied up today, but when I get
> the chance I'll pursue that

I have had some trouble a while ago with a particular usb card reader that
if plugged in on boot up but with no CF or SD card installed would cause
the system to hang. If it had a card in either slot or the reader just not
plugged
in then the system would boot. That was even with auto mount of the media
turned off in fstab. I just boot with it removed on that machine now and
connect
on the very rare case of needing to do so. That was using 2005LE I've
never
tested it on boot through the last few changes to Mandriva, I use it so
rarely.
This issue appeared just after smbd and nmbd were trying to run, your
mileage
may differ but it could be a similar issue.

However, I would be looking to see why your SSH connection is dropping first
as that at least gives you control of your machine without the need to hit
the
power button. A cron job that is running on a very long schedule?

Andrew Ed