Have you downloaded the MIB and searched through it? Also, why?
IMNSHO one should be synchronizing the time on your network devices with a stable NTP time source; in many organizations domain controllers make a good source, in others a Unix NTP servers makes sense. [a good set of NTP reference servers can be found at USNO Network Time Servers or pool.ntp.org: the internet cluster of ntp servers]
Then it's only a case of looking for messages that they have lost peering or made large time adjustments.
Thank you for your reply.
As you probably know there is not "a" MIB for complex switches but quite a few.
All I was able to find so far is the date of the last reboot for example, but not the current time stamp on the device.
While it is probably true that in smaller networks one can be confident that a network device is not suddenly cut off from
an NTP server by either vLAN changes, routing or some other issues, the same does not apply to large networks with
hundreds of switches, firewall, etc.
So, in order to find a device which clock is off I am polling device time via SNMP. If anyone knows the OIDs for date and
time for HP switches that would be appreciated.
1 of 1 people found this helpful
Unfortunately, it is not possible to get current date/time from switch via SNMP.
Instead of monitoring switches for NTP, try to monitor links to NTP Server.
I suppose you need this for syslog + maybe some timestaps for configs.
Correct me if Im wrong.
Is there any other "services" on switch, that actual date/time is required?
Thank you for your reply.
It is mostly about being able to look into a devices log file to see what happened when
something went wrong and to have the correct time stamp in the log.
Some older devices on our network don't implement daylight savings at all or at least
not correctly, so the only way to monitor the current time stamp is by polling the devices
clock. Some devices don't even support clock updates via NTP at all.
I find it curious that all of our Dell PowerConnect switches let you do this via SNMP, but
none of the much more expensive ProCurves seem to let you do this.
Oh well, thanks for confirming what I already suspected.
I hope that our Extreme switches will let me poll time and date.
I manage a moderate sized network of 4000+ switches, 5000+ wireless APs, and 500+ routers. The standard MIB for the systemDate is is hrSystemDate; I've looked through the HP mib and there's nothing obvious I can see that is custom there. I'd ask HP to support the standard Host Resource MIB since that also reports many standard useful things.
How we ensure time is synchronized:
- timeservers are on the same/diverse subnets as DNS servers, radius servers, domain controllers; if all of them are not reachable then we have a more significant network problem than not being able to set the clock on a switch.
- All of our switches (and other network devices) syslog and trap to multiple logservers (Solarwinds Orion primary and secondary clusters being two of them)
- NCM's configuration policy reporting to ensure trap, time, and syslog servers are configured correctly across the network
- Custom SQL alerts in the advanced alert manager to generate alerts if the time servers cannot be reached based on the syslog/trap messages (which are more likely to make it through the network as one-way messages than a request/response protocol)
for reference, sampling switches here for RESOURCES-MIB::hrSystemDate.0 support:
- HP (Procurve 26xx, 4xxx, 8xxx, 54xx) : No Such Object available on this agent at this OID
- Juniper (all EX-series) : HOST-RESOURCES-MIB::hrSystemDate.0 = STRING: 2013-9-4,8:44:42.0,-7:0
- Brocade:No Such Object available on this agent at this OID
- Extreme: No Such Object available on this agent at this OID
- D-Link: No Such Object available on this agent at this OID
- SMC: No Such Object available on this agent at this OID
Thanks for the info Richard!