Test NeDi 1.4 Patch5

[rickli]

Here it is. Please post your findings. I'll make it public, if all your issues are fixed...

http://www.nedi.ch/pub/nedi-1.4p5.tgz

[ascii]

everythink is working for me.
devices can be added to monitoring again

[jacobl]

I've run into some errors and unexpected behavior doing a discovery after re-initializing the database. I recall not having these issues on a virgin install (NeDiO ISO + patch3). Not quite ready to do a detailed bug report as I need to narrow it down to the specific patch or configuration setting, which I will try to do in the next few days.

[Kasper]

I have run this patch for about a week and not noticed any problems.
I'm not the most advanced NeDi user but I thought I should at least report back the success for us at least.

[rickli]

Such feedback is always welcome

[sjobergh]

When using Nodes,List,  adding column "user" in the scrollbox I dont get the username,  I get numerical value in return.
Looking in the database I can see the user in alpha

[Hannu Liljemark]

1.4p5 looking pretty good here.

Is anyone else running it somewhere with PoE devices? I have so far one instance where the "Total PoE" graph isn't graphing any more after going from 1.0.9 to 1.4p5 (environment with mainly Cisco switches), and two instances where the total poe graph is ok after upgrade (Procurve switches).

nedi.conf has "usepoe ifmib" in all cases.

Br,
Hannu

[Grisu]

The mysql deadlock errors are still present with Patch 5 

Try restarting transaction at /var/nedi/inc/libdb.pm line 1874
Try restarting transaction at /var/nedi/inc/libdb.pm line 1541

[Hannu Liljemark]

FYI, deleting a node deletes the node but last line is this:

Unknown column 'mac' in 'where clause'

Br,
Hannu

[rickli]

Tx, Hannu. The unknown column is ugly, but can be ignored (it's marked for Patch6 should there be a need for it). 1.4 support the Power-Ethernet Mib, which is used for total power consumption rather than adding all interfaces as 1.0.9 did. Check the .def where this is not showing...

Grisu, do you have any parallel discoveries?

[Hannu Liljemark]

1.4 support the Power-Ethernet Mib, which is used for total power consumption rather than adding all interfaces as 1.0.9 did. Check the .def where this is not showing...

Thanks, figured out this def issue on my own yesterday after started poking around, before seeing your response. Feels like deja vu... I think I've done something like this before. Anyway, I had to modify these definitions:

1.3.6.1.4.1.11.2.3.7.11.55.def, IFpowr  1.3.6.1.4.1.11.2.14.11.1.9.1.1.1.3.1    P
1.3.6.1.4.1.11.2.3.7.11.155.def, IFpowr  1.3.6.1.4.1.11.2.14.11.1.9.1.1.1.3.1    P
1.3.6.1.4.1.9.1.951.def, IFpowr  1.3.6.1.4.1.9.9.402.1.2.1.7     P
1.3.6.1.4.1.9.1.950.def, IFpowr  1.3.6.1.4.1.9.9.402.1.2.1.7     P
1.3.6.1.4.1.9.1.1016.def, IFpowr  1.3.6.1.4.1.9.9.402.1.2.1.7     P
1.3.6.1.4.1.9.1.1830.def, IFpowr  1.3.6.1.4.1.9.9.402.1.2.1.9     P

I'm not sure if it matters on Cisco devices whether you use 1.3.6.1.4.1.9.9.402.1.2.1.9 (consumption) or 1.3.6.1.4.1.9.9.402.1.2.1.7 (allocated).

Br,
Hannu

[rickli]

Tx, added the missing info. I'd go for whatever is closer to real life, so consumption?

[Hannu Liljemark]

Tx, added the missing info. I'd go for whatever is closer to real life, so consumption?

Real life value would give you accurate number of how much power is drawn by PoE devices. I wonder how e.g. a Cisco switch works regarding the allocated value - if connected devices draw power 100W but switch has allocated 200W, you probably can't over-subscribe the switch or can you.

I recall an instance with 48p switch that had maybe 20-30 phones and next phone didn't get power as we hit the switch PSU's limit. Anyway, just wondering if people use Nedi PoE data to see if their switches still have room for new PoE clients... or maybe this doesn't matter at all - Nedi is your creation and you set the rules

Right now the defs seem to be mixing 1.3.6.1.4.1.9.9.402.1.2.1.9 (consumption) and 1.3.6.1.4.1.9.9.402.1.2.1.7 (allocated).

Br,
Hannu

[rickli]

People using NeDi should get the most useful data out of it, thus we all decide However a mix is definitely not what we want, so it boils down to the question whether the switch acts upon real or allocated values. I can change them all to 1.3.6.1.4.1.9.9.402.1.2.1.9 until further enlightenment...

[Hannu Liljemark]

However a mix is definitely not what we want, so it boils down to the question whether the switch acts upon real or allocated values. I can change them all to 1.3.6.1.4.1.9.9.402.1.2.1.9 until further enlightenment...

Agreed!

Two more needed modifications:

1.3.6.1.4.1.9.1.950.def, IFpowr  1.3.6.1.4.1.9.9.402.1.2.1.9
1.3.6.1.4.1.9.1.1229.def, IFpowr  1.3.6.1.4.1.9.9.402.1.2.1.9

I also noticed that there's 1.3.6.1.4.1.9.9.402.1.2.1.7.def file in 1.4 for WS-C3750V2-24PS-S. I wonder if the filename is a typo (as it's the oid for this poe stuff). Looking at http://tools.cisco.com/Support/SNMP/do/BrowseOID.do?objectInput=1.3.6.1.4.1.9.1&translate=Translate&submitValue=SUBMIT&submitClicked=true the correct type oid for WS-C3750V2-24PS-S might be 1.3.6.1.4.1.9.9.1023.

Br,
Hannu