Send netdisco-users mailing list submissions to
[email protected]
To subscribe or unsubscribe via the World Wide Web, visit
https://lists.sourceforge.net/lists/listinfo/netdisco-users
or, via email, send a message with subject or body 'help' to
[email protected]
You can reach the person managing the list at
[email protected]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of netdisco-users digest..."
Today's Topics:
1. Re: Netdisco2 sometimes misses connected hosts data & SSH
Collector (Muris)
2. Re: Netdisco2 sometimes misses connected hosts data & SSH
Collector (Oliver Gorwits)
3. Re: Netdisco2 sometimes misses connected hosts data & SSH
Collector (Muris)
--- Begin Message ---
Hi Oliver,
Thanks for replying.
Here is my exact problem as I have spent 2 days analysing it, as it took me
ages to figure out whats happening with the large amount of devices, lets say I
have this setup at the moment:
Switch1
CoreSwitch1
Between switches its setup as L2 with port channels.
Now for example..
1)When I run a arpnip/macsuck on Switch1, everything appears fine.
2)When it comes along to run a macsuck/arpnip on CoreSwitch1, the
arpnip/macsuck info disappears off the ports from Switch1, but instead shows up
on the portchannel on CoreSwitch1 that links to switch1 with all the hosts from
Switch1.
3)Then when the next arpnip/macsuck is done on Switch1, all the data appears on
Switch1 ports and disappears off the CoreSwitch1 port channel. Its almost like
one moment its on one switch, next moment its on the next, keeps swapping
backwards and to when each arpnip/mac suck is probed.
So all mac/arp comes on the port channel on core switch, then released, on,
then released , depending when each probe is done. Switch1 shows data on and
off on each probe.
If all the data is appearing correctly on Switch1 on all the ports, how can I
prevent the uplink on CoreSwitch1 show all the arp/mac when its polled. Ive
been trying for hours looking through the documentation how to prevent this
occurring, but didn’t find anything.
Its almost like I need something to specify its an uplink, don’t add any
macs/arps on this port..or I could be just missing something else im not seeing.
Thanks Muris
On 24/6/17, 6:57 pm, "Oliver Gorwits" <[email protected]> wrote:
Hi Muris
Please can you describe in a little more detail what's missing from the
web interface - is it the MAC addresses of nodes on the ports, or the IP
addresses of those nodes, or the neighbor relations to other devices?
Each of these is gathered by a different process.
You could try slowing down the pollers by setting sleep_time:
workers:
tasks: 'AUTO * 4'
sleep_time: 2
This makes each worker pause a little after it polls (the default is one
second). If your problem is UDP overload on your server then it might
help.
regards,
oliver.
On 2017-06-23 11:37, Muris wrote:
> Hi All,
>
> I would like to say thanks to the person who came up with the SSH
> Collector for Nexus/ASA devices its really awesome and working well.
> Can it also connect via telnet ?
>
> I have come up with a problem, as i have around 3000 devices, for some
> reason the macksuck and arp, on the automatic arpnips/walks dont fully
> populate the data for a switch, when i go into the switch some ports
> show up but blanks next to them on connected devices..
>
> However, if i manually force a discover/arpnip/macwalk through the
> webinterface the data populates straight away.
>
> Is there any reason for this? How can it completely do it automated
> without me refreshing a device manually? (Well the netdisco-backend
> should be doing it but some reason some data dissapears when
> displayed)
>
> I was thinking the SSH Collector of arps maybe could be conflicting
> with netdisco arps to populate data...not sure..but at the moment
> needing the ssh collector due to mpls and nexus with vrfs etc.
>
> To cater for this many devices, i have 8 CPUs set on a Xeon machine
> with 16gb ram, and running the backend worker on AUTO * 4 in
> deployment config.
>
> Postgresql is optimised with pgtune, and running latest netdisco
> version released 22/06/17
>
> Any troubleshooting/recommendations would be appreciated :)
>
> Thankyou, Muris
>
>
>
------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>
> _______________________________________________
> Netdisco mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/netdisco-users
--- End Message ---
--- Begin Message ---
Hi Muris
Thanks for the detailed response.
This can happen when you have macsuck_bleed set to true in your config -
so please check for that.
Otherwise, when the nodes all move to the port channel on CoreSwitch1,
is there a known neighbor on that port channel? (that is, does Netdisco
report that Switch1 is connected on the port).
Is Netdisco reporting "possible device uplink" on either end of the port
channel?
regards,
oliver.
On 2017-06-24 19:34, Muris wrote:
Hi Oliver,
Thanks for replying.
Here is my exact problem as I have spent 2 days analysing it, as it
took me ages to figure out whats happening with the large amount of
devices, lets say I have this setup at the moment:
Switch1
CoreSwitch1
Between switches its setup as L2 with port channels.
Now for example..
1)When I run a arpnip/macsuck on Switch1, everything appears fine.
2)When it comes along to run a macsuck/arpnip on CoreSwitch1, the
arpnip/macsuck info disappears off the ports from Switch1, but instead
shows up on the portchannel on CoreSwitch1 that links to switch1 with
all the hosts from Switch1.
3)Then when the next arpnip/macsuck is done on Switch1, all the data
appears on Switch1 ports and disappears off the CoreSwitch1 port
channel. Its almost like one moment its on one switch, next moment its
on the next, keeps swapping backwards and to when each arpnip/mac suck
is probed.
So all mac/arp comes on the port channel on core switch, then
released, on, then released , depending when each probe is done.
Switch1 shows data on and off on each probe.
If all the data is appearing correctly on Switch1 on all the ports,
how can I prevent the uplink on CoreSwitch1 show all the arp/mac when
its polled. Ive been trying for hours looking through the
documentation how to prevent this occurring, but didn’t find anything.
Its almost like I need something to specify its an uplink, don’t add
any macs/arps on this port..or I could be just missing something else
im not seeing.
Thanks Muris
On 24/6/17, 6:57 pm, "Oliver Gorwits" <[email protected]> wrote:
Hi Muris
Please can you describe in a little more detail what's missing from
the
web interface - is it the MAC addresses of nodes on the ports, or
the IP
addresses of those nodes, or the neighbor relations to other
devices?
Each of these is gathered by a different process.
You could try slowing down the pollers by setting sleep_time:
workers:
tasks: 'AUTO * 4'
sleep_time: 2
This makes each worker pause a little after it polls (the default
is one
second). If your problem is UDP overload on your server then it
might
help.
regards,
oliver.
On 2017-06-23 11:37, Muris wrote:
> Hi All,
>
> I would like to say thanks to the person who came up with the SSH
> Collector for Nexus/ASA devices its really awesome and working
well.
> Can it also connect via telnet ?
>
> I have come up with a problem, as i have around 3000 devices, for
some
> reason the macksuck and arp, on the automatic arpnips/walks dont
fully
> populate the data for a switch, when i go into the switch some
ports
> show up but blanks next to them on connected devices..
>
> However, if i manually force a discover/arpnip/macwalk through
the
> webinterface the data populates straight away.
>
> Is there any reason for this? How can it completely do it
automated
> without me refreshing a device manually? (Well the
netdisco-backend
> should be doing it but some reason some data dissapears when
> displayed)
>
> I was thinking the SSH Collector of arps maybe could be
conflicting
> with netdisco arps to populate data...not sure..but at the moment
> needing the ssh collector due to mpls and nexus with vrfs etc.
>
> To cater for this many devices, i have 8 CPUs set on a Xeon
machine
> with 16gb ram, and running the backend worker on AUTO * 4 in
> deployment config.
>
> Postgresql is optimised with pgtune, and running latest netdisco
> version released 22/06/17
>
> Any troubleshooting/recommendations would be appreciated :)
>
> Thankyou, Muris
>
>
>
------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>
> _______________________________________________
> Netdisco mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/netdisco-users
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Netdisco mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/netdisco-users
--- End Message ---
--- Begin Message ---
Hi Oliver, thanks again for looking into.
There is no mac_suck bleed set at all, i cannot see anything with that in
deployment config.
Sometimes it reports possible uplink, and sometimes it doesnt.
Its not vendor specific either.. for example i have these scenarios also
happening. All the interconnectivity is L2 trunks
HP Comware Switch1 <> HP Comware Core Switch <> HP Comware Switch2
If the hosts sit on Switch1, sometimes i can see them on the port
channel/Bridge Aggregation on switch2's uplink as well (but not on core
switch). If i do a mac/arp on Switch1 they re-appear correctly there on the
ports and things are fine for a little while, then when the process does a
mack suck on switch 2, things get taken off Switch1, then put on the port
channel on Switch2. Things kind of cycle like this between the devices and
also as ive described above, which makes it difficult to see when finding
where a hosts are connected as it keeps floating.
I also have the same problem with Cisco 3850 switches and L2 trunks
L3 Device Nexus 7k << L2 Trunk>> 2x 3850 stack <<<L2 Trunk>>> 2 x 3850
stack <<L2 Trunk>> Nexus 7k L3 Device
hosts can float on the trunks back and forth from user ports, so one minute
you looking up a device it appears correctly on a switchport, and you
check an hour later, its gone off that switchport and appearing on one of
the trunks. Again sometimes these appear as no uplink, or maybe uplink.
I would be so happy if this can be rectified in some way. Maybe an option
to override and specify its definite uplink dont accumulate macsuck/arpnip
data on the specific problematic uplink, but this could be a administrative
issue especially in dealing with thousands of devices..be easier having it
automatically know its certainly an uplink from lldp/cdp.
All CDP/LLDP is enabled on the network however.
With having blade chassis/servers connected to switches I would like to
know whats on the port channels to these macs/arps etc, so it would need to
differentiate between a switch and something else, if automatic fix took
place to filter data on uplinks.
As it occurs accross multiple platforms, it must be something with the
discovery process that its not getting correctly.
Any help as always is greatly appreciated so i can have it working well.
Thankyou, Muris
On Sun, Jun 25, 2017 at 4:04 AM, Muris <[email protected]> wrote:
> Hi Oliver,
>
> Thanks for replying.
>
> Here is my exact problem as I have spent 2 days analysing it, as it took
> me ages to figure out whats happening with the large amount of devices,
> lets say I have this setup at the moment:
>
> Switch1
> CoreSwitch1
>
> Between switches its setup as L2 with port channels.
>
> Now for example..
>
> 1)When I run a arpnip/macsuck on Switch1, everything appears fine.
> 2)When it comes along to run a macsuck/arpnip on CoreSwitch1, the
> arpnip/macsuck info disappears off the ports from Switch1, but instead
> shows up on the portchannel on CoreSwitch1 that links to switch1 with all
> the hosts from Switch1.
> 3)Then when the next arpnip/macsuck is done on Switch1, all the data
> appears on Switch1 ports and disappears off the CoreSwitch1 port channel.
> Its almost like one moment its on one switch, next moment its on the next,
> keeps swapping backwards and to when each arpnip/mac suck is probed.
> So all mac/arp comes on the port channel on core switch, then released,
> on, then released , depending when each probe is done. Switch1 shows data
> on and off on each probe.
>
> If all the data is appearing correctly on Switch1 on all the ports, how
> can I prevent the uplink on CoreSwitch1 show all the arp/mac when its
> polled. Ive been trying for hours looking through the documentation how to
> prevent this occurring, but didn’t find anything.
>
> Its almost like I need something to specify its an uplink, don’t add any
> macs/arps on this port..or I could be just missing something else im not
> seeing.
>
> Thanks Muris
>
>
> On 24/6/17, 6:57 pm, "Oliver Gorwits" <[email protected]> wrote:
>
> Hi Muris
>
> Please can you describe in a little more detail what's missing from the
> web interface - is it the MAC addresses of nodes on the ports, or the
> IP
> addresses of those nodes, or the neighbor relations to other devices?
> Each of these is gathered by a different process.
>
> You could try slowing down the pollers by setting sleep_time:
>
> workers:
> tasks: 'AUTO * 4'
> sleep_time: 2
>
> This makes each worker pause a little after it polls (the default is
> one
> second). If your problem is UDP overload on your server then it might
> help.
>
> regards,
> oliver.
>
> On 2017-06-23 11:37, Muris wrote:
> > Hi All,
> >
> > I would like to say thanks to the person who came up with the SSH
> > Collector for Nexus/ASA devices its really awesome and working well.
> > Can it also connect via telnet ?
> >
> > I have come up with a problem, as i have around 3000 devices, for
> some
> > reason the macksuck and arp, on the automatic arpnips/walks dont
> fully
> > populate the data for a switch, when i go into the switch some ports
> > show up but blanks next to them on connected devices..
> >
> > However, if i manually force a discover/arpnip/macwalk through the
> > webinterface the data populates straight away.
> >
> > Is there any reason for this? How can it completely do it automated
> > without me refreshing a device manually? (Well the netdisco-backend
> > should be doing it but some reason some data dissapears when
> > displayed)
> >
> > I was thinking the SSH Collector of arps maybe could be conflicting
> > with netdisco arps to populate data...not sure..but at the moment
> > needing the ssh collector due to mpls and nexus with vrfs etc.
> >
> > To cater for this many devices, i have 8 CPUs set on a Xeon machine
> > with 16gb ram, and running the backend worker on AUTO * 4 in
> > deployment config.
> >
> > Postgresql is optimised with pgtune, and running latest netdisco
> > version released 22/06/17
> >
> > Any troubleshooting/recommendations would be appreciated :)
> >
> > Thankyou, Muris
> >
> >
> > ------------------------------------------------------------
> ------------------
> > Check out the vibrant tech community on one of the world's most
> > engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> >
> > _______________________________________________
> > Netdisco mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/netdisco-users
>
>
>
>
>
>
--- End Message ---
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Netdisco mailing list - Digest Mode
[email protected]
https://lists.sourceforge.net/lists/listinfo/netdisco-users
