Constantly lost network connection to hp-ux server

[Andrey Deryugin]

Hello,

I've installed HP-UX 11i v1 december 2003 on our rp3410 server.
After configure the network I can reach server via X, ftp, telnet etc. and also I can access any external hosts as well. But periodically server became unreachable. Even can't ping it. When I ping it from my Windows machine sometimes I get 1 reply and after I get 'Request timed out', some times its just request timed out.
I tried reconnect it to different Cisco switch, but it doesn't help.

Can somebody point me what I did wrong?

Thanks in advance,
Andrey

P.S.
I didn't apply any patches so far.

[harry d brown jr]

Have you tried to replace the cables? Also, make sure the card is properly seeded and isn't awry.

When you "can't reach" the server, try using the IP instead of the name. If that works consistantly, then it's probably a DNS issue.

Also, check the syslog to see if you have any LAN failures.

live free or die
harry

[RAC_1 · Accepted solution]

I think you ahve problem of dead gateway setting.

ndd -h ip_ire_gw_probe
ndd -get /dev/ip ip_ire_gw_probe
The default is 1.
If set to 1, set it to 0

ndd -set /dev/ip ip_ire_gw_probe 0

for permanent changes, put it in /etc/rc.config.d/nddconf file.

Other than this, check cable connections, lan cables etc.

Anil

[Andrey Deryugin]

It seems like that Anil is right.
Because if I ping gateway from HP-UX RS232 console the connection restores immediately.

But thing is I just set it to 0 and get it again :(

One more question about set it in nddconf.

Seems I have to set it like:

TRANSPORT_NAME[0]=ip
NDD_NAME[0]=ip_ire_gw_probe_interval
NDD_VALUE[0]=0

I'm not sure about number in square brackets - is that just a record number?

Regards,
Andrey

[Jeff Schussele]

Hi Andrey,

It's an index number & *must* be unique for each entry. Just increment it by one for each new entry.
Oh & just setting it in the file does nothing until the next boot - so run

ndd -set /dev/ip ip_ire_gw_probe 0

That will set it immediately. The entry in nddconf sets it at every subsequent boot.

Rgds,
Jeff

[Bharat Katkar]

Hi,
that is index number and need's to be incremented as you go on adding additional tunable parameter. Remember you have to maintain the sequence.

Reference: man ndd, more nddconf

Use
# ndd -c
Makes the configuration in nddconf to take effect immediately.

Regards,

[Andrey Deryugin]

Hello Jeff,

Yes, I did set it as Alin recommended, but lost connection 3 times already after that.

Also according to ndd -h ip_ire_gw_probe it's not a switch on/off, but "the probe interval for Dead Gateway Detection. ... With retries, the maximum time to detect a dead gateway is ip_ire_gw_probe_interval + 10000 milliseconds ...".

IS there any possibility to switch off the probe itself?

Regards,
Andrey

[Jeff Schussele]

Hi (again) Andrey,

You're confusing ip_ire_gw_probe - which IS the proper value - with ip_ire_gw_probe_interval which sets the time value IF ip_ire_gw_probe=1.
If it equals 0 then the probe is entirely disabled.

Rgds,
Jeff

[Andrey Deryugin]

Hmm,

ndd -c caused "operation failed, ip_ire_gw_probe_interval: errno = 22"

Regards,
Andrey

[Andrey Deryugin]

Thanks Jeff,

That's 'ndd -h supp' confused me, because ip_ire_gw_probe doesn't exist in that list
and 'ndd -h ip_ire_gw_probe' shows me help for 'ip_ire_gw_probe_interval' :)

Thanks everyone!

Regards,
Andrey

[RAC_1]

Which OS???

Anil

[Andrey Deryugin]

uname -a

HP-UX rp3410 B.11.11 U 9000/800 2848620217 unlimited-user license

The bad thing is it doesn't help much.
Still lost connection.

Regards,
Andrey

[RAC_1]

Execute only the following command and post.

ndd -set /dev/ip ip_ire_gw_probe 0

Anil

[harry d brown jr]

I have never been forced to mess with the ip_ire_gw_probe settings or most of the other ndd settings on 11i.

Can you post the output from:

ioscan -fn -C lan

lanscan -v

and a lanadmin on each ppa

Again, I think you should try replacing the cables and possibly moving to another port on your cisco switch.

live free or die
harry

[Andrey Deryugin]

# ioscan -fn -C lan
Class I H/W Path Driver S/W State H/W Type Description
===================================================================
lan 0 0/1/2/0 igelan CLAIMED INTERFACE HP PCI 1000Base-T Core

# lanscan -v
-------------------------------------------------------------------------------
Hardware Station Crd Hdw Net-Interface NM MAC HP-DLPI DLPI
Path Address In# State NamePPA ID Type Support Mjr#
0/1/2/0 0x00306E4B0D9D 0 UP lan0 snap0 1 ETHER Yes 119

Extended Station LLC Encapsulation
Address Methods
0x00306E4B0D9D IEEE HPEXTIEEE SNAP ETHER NOVELL

Driver Specific Information
igelan
-------------------------------------------------------------------------------

LAN INTERFACE STATUS DISPLAY
Fri, Jul 23,2004 18:55:11

PPA Number = 0
Description = lan0 HP PCI Core I/O 1000Base-T Release B.11.1
1.11
Type (value) = ethernet-csmacd(6)
MTU Size = 1500
Speed = 10000000
Station Address = 0x306e4b0d9d
Administration Status (value) = up(1)
Operation Status (value) = up(1)
Last Change = 1010081
Inbound Octets = 722566953
Inbound Unicast Packets = 1003477
Inbound Non-Unicast Packets = 39863
Inbound Discards = 0
Inbound Errors = 0
Inbound Unknown Protocols = 10340
Outbound Octets = 69281110
Outbound Unicast Packets = 414576
Outbound Non-Unicast Packets = 24
Outbound Discards = 0
Outbound Errors = 0
Outbound Queue Length = 0
Specific = 655367

Ethernet-like Statistics Group

Index = 1
Alignment Errors = 188678
FCS Errors = 0
Single Collision Frames = 42799
Multiple Collision Frames = 91706
Deferred Transmissions = 139728
Late Collisions = 0
Excessive Collisions = 0
Internal MAC Transmit Errors = 0
Carrier Sense Errors = 0
Frames Too Long = 0
Internal MAC Receive Errors = 4294584753

I did change cable and I did 'ndd -set /dev/ip ip_ire_gw_probe 0'

After 1 hour everything is ok.

Regards,
Andrey

[harry d brown jr]

The sad thing is that you have a GigE card and it is running at 10Megabits (speed=10,000,000).

Can your cisco switch handle GigE or 100BT ??

live free or die
harry

[Andrey Deryugin]

It's a Catalyst 1900 with 24 10BaseT and 2 100BaseT ports.
That's not so important for now. Later I will reconnect it.

I gonna leave it with X client connected for weekend, lets see what happend.

Thanks again for the help.

Regards,
Andrey

[Dan Maul]

I had same problem and it wasn't until I did a ping -a IP did I detect the problem. Someone else was using my IP address. Could he your problem as well.

Good luck.

[Andrey Deryugin]

Thanks Dan, that was it!

Andrey