SRX

Here's a Cacti host template that I put together, tested on SRX100, started with a smaller template for an EX series so it should be generic enough to work on most models. It includes CPU, memory, flow session count, storage used, and system temperature. The only thing I'm not sure of is the storage layout being the same across devices, there weren't labels for the SNMP indexes, so you may have to change the last digit on the OIDs for the various data templates. Other than that, have fun, and please, if you add to it post it back.

Note that I changed the extension to .txt to satisfy the forum's attachment requirements.

 

Thanks to someITguy for pointing out that some of the OIDs were missing '.' prefixes. Apparently cacti was tolerant of this in many cases, but it was obviously a mistake and didn't work for his installation.

thanks for this! Please reply to this so I can credit you a 'solved' thread.

 

-Keith

No problem, just hope people will find it useful.

I was just last week testing cacti and wondering if there were any SRX plugins... This seems to work with SRX650 just fine, many thanks!

Hi there,

 

Thanks for the template, but when I try importing your template into Cacti I get this error message - Error: XML: hash version does not exist.

 

I am using Cacti Version 0.8.7b

 

Any help would be highly appriciated.

 

Thanks

 

 

take a look at this:

 

http://docs.cacti.net/howto:determine_cacti_template_version

 

basically you'll have to be running 0.8.7e or newer.  Sorry, you may have to upgrade!  

Thanks for letting me know I will try that.

 

Cheers

 

---

@obi-lan wrote:

I was just last week testing cacti and wondering if there were any SRX plugins... This seems to work with SRX650 just fine, many thanks!

---

 

Hi obi-lan,

with oure SRX650 the CPU and Memory Chart wont work. With OID do you use for the two values in the SRX?

 

With the SRX100 the template works fine! Many thanks for that!

 

Best Regards

Jochen

I can't help you with 650 specifics, but I can perhaps give you the oids to help you find it.  Do you have the mibs installed? If so, walk JUNIPER-MIB::jnxOperatingTable and see if any of them fit.

snmpwalk -v2c -c @readonly 10.10.1.1 JUNIPER-MIB::jnxOperatingTable

 

If not, that corresponds to oid .1.3.6.1.4.1.2636.3.1.13, but you may have trouble deciphering what's what. Here's what I get for descriptions and corresponding CPU util:

 

 

SNMPv2-SMI::enterprises.2636.3.1.13.1.5.1.1.0.0 = STRING: "midplane"
SNMPv2-SMI::enterprises.2636.3.1.13.1.5.2.1.0.0 = STRING: "PEM 0"
SNMPv2-SMI::enterprises.2636.3.1.13.1.5.7.1.0.0 = STRING: "FPC: FPC @ 0/*/*"
SNMPv2-SMI::enterprises.2636.3.1.13.1.5.8.1.1.0 = STRING: "PIC: 8x FE Base PIC @ 0/0/*"
SNMPv2-SMI::enterprises.2636.3.1.13.1.5.9.1.0.0 = STRING: "Routing Engine"
SNMPv2-SMI::enterprises.2636.3.1.13.1.5.9.1.1.0 = STRING: "USB Hub"

SNMPv2-SMI::enterprises.2636.3.1.13.1.8.1.1.0.0 = Gauge32: 0
SNMPv2-SMI::enterprises.2636.3.1.13.1.8.2.1.0.0 = Gauge32: 0
SNMPv2-SMI::enterprises.2636.3.1.13.1.8.7.1.0.0 = Gauge32: 0
SNMPv2-SMI::enterprises.2636.3.1.13.1.8.8.1.1.0 = Gauge32: 0
SNMPv2-SMI::enterprises.2636.3.1.13.1.8.9.1.0.0 = Gauge32: 20
SNMPv2-SMI::enterprises.2636.3.1.13.1.8.9.1.1.0 = Gauge32: 0

If you're not familiar with the way it indexes, it does so off of the last four digits, so '9100' refers to my routing engine, and then each OID ending in 9100 is a stat of that routing engine. So maybe your routing engine just has a different index.  Memory is the same:

 

SNMPv2-SMI::enterprises.2636.3.1.13.1.11.9.1.0.0 = Gauge32: 60

and temperature:

SNMPv2-SMI::enterprises.2636.3.1.13.1.7.9.1.0.0 = Gauge32: 49

I'd recommend installing or at least reading the mib files because they'll give you info on what the values mean, what units they're in, etc.

Thanks for your reply sor.

 

I know all that information about OIDs and I already have the MIB Tables.

There I found no way to monitor the CPU for the SRX650.

I ask the support and the told me, that the SRX650 still not support monitor the CPU via SNMP.

 

So I will wait to the next JUNOS versions and hope, that this "feature" will come soon 😉

Smarthouse_JB:

 

I use the template without modification and did work straight. I think the OID is .1.3.6.1.4.1.2636.3.1.13.1.8.9.1.0.0 for CPU. Same worked on SRX240 too. Software version on  SRX650 is 10.1R1.8

Hej obi-lan

Ok, i use the version 10.0.R2.10 on the SRX650 Cluster.

On this version the OID .1.3.6.1.4.1.2636.3.1.13.1.8.9.1.0.0 doesen't exist.

 

On the other platforms it works with this version pretty good.

 

Cheers Jochen

There's a known issue with SRX clusters in current code where jnxOperatingTable and, more specifically, the jnxBoxAnatomy parent aren't populated.  I've had a case open on this for a while and the issue has been acknowledged and sent to engineering though I'm still waiting on an official PR.

None of the template values seem to be working for me.. Normal SNMP data such as the interfaces seem to be fine and I don't seem to be getting errors..

 

Is there any specific configuration needed? SNMP version? or any host specific stuff?

 

Right now I simply have a an SNMP allow for the Cacti host and SNMP v2 set in cacti.

To clarify, what do you mean by 'template isn't working, normal SNMP data works', the cacti template works for you but only on interfaces, not on the indexed items (disk space)? Or it works when you manually add the host but the template doesn't work?

Sounds as though you're asking about setting up SNMP perms, so you probably aren't even sure if you can get SNMP data back yet?

I'll ask, are you able to snmpwalk the SRX from your cacti host? If so, then you probably need to tweak some of the OIDs in the template. Use snmpwalk and the MIBs from the juniper website to determine what you're after.  If not, then you probably need to focus on proving that the SRX is accepting/returning queries, even independent of cacti.

 

The template is really just a grouping of prelabeled OIDs to query and graph definitions, so it should be transparent of anything else, so long as you can 1) import the template into cacti, and 2) query the SRX from the cacti host via SNMP and return the OID data.

I know how templates work, what I am saying is that none of the special data-sources from this template graph on the SRX 210 I am testing against runing JunOS 10.1R1.8.

 

I haven't had a chance to rip apart the template and walk the devices.

 

What does work is the generic interface data-sources. IE graphing bandwidth on ge-0/0/1 works just fine so I know SNMP is working.

 

What I was asking before diging deeper was if a specific SNMP version was needed to be set in Cacti (currently set to SNMP v2) or if any special settings where needed on the SRX to export of graph these extra data points.

To answer that question, no. Templates in general are transparent to the SNMP configuration of a device, whether you choose v2 or v3 auth, encryption type, etc , so long as cacti can pull data with the credentials it's given.

 

If you can get the interfaces through the generic templates, then it's strange that the interface templates in this host template don't work, because they should be the exact same data template. I just included the standard interface data template, into the host template, and the graph template for that was only slightly modified.  If it weren't for that, I would suspect that you were only exporting specific OIDs to the community you're querying.

 

You shouldn't have to rip apart the template to walk the device. See the OID posted earlier, just walk that and it should return the juniper chassis specific stuff. That will at least show you if it's capable of retrieving CPU/mem/temp.

 

Ah, something else to check is to look and see if the data is making it to the RRD. Or better yet, just look at the graph and enable debug, see if the RRDs/data are there.  I've seen issues with cacti in the past where I've had to manually adjust the data sources being polled for a particular graph, even though all the data was there (mainly multicore cpu stuff).

 

---

@sor wrote:

  If it weren't for that, I would suspect that you were only exporting specific OIDs to the community you're querying.

 

---

 

Ok, then maybe my SNMP setting on the SRX is wrong, this is what I have where x values have been blocked out for privacy.

 

 

snmp {
    community xxxxxxxxxx {
        authorization read-only;
        clients {
            10.x.x.x/32;
        }
    }
    health-monitor;
}

 

 

How did you made this working. Can you please help on giving some steps how to do so.

 

-Arien

I have also walked the srx from my cacti box and although a lot is returned I do not get anything for the OID in the template

Ok, so you have health-monitor on. What do you get when you type 'show snmp health-monitor' on the cli?  For me, this shows the OIDs that are in the template... translated per the MIB definitions, i.e. the '9.1.0.0' cpu, etc.

 

root@cerberus> show snmp health-monitor    

Alarm 
Index  Variable description                           Value State

 32768 Health Monitor: root file system utilization
       jnxHrStoragePercentUsed.1                         64 active              

 32769 Health Monitor: /config file system utilization
       jnxHrStoragePercentUsed.2                          0 active              

 32770 Health Monitor: RE 0 CPU utilization
       jnxOperatingCPU.9.1.0.0                           93 rising threshold    

 32771 Health Monitor: RE 1 CPU utilization
       jnxOperatingCPU.9.2.0.0                            0 object not available

 32772 Health Monitor: RE 0 memory utilization
       jnxOperatingBuffer.9.1.0.0                        61 active              

 32773 Health Monitor: RE 1 memory utilization
       jnxOperatingBuffer.9.2.0.0                         0 object not available

 32774 Health Monitor: Max Kernel Memory Used (%) 
       jnxBoxKernelMemoryUsedPercent.0                    1 active              

Played around with GetIf a little and confirmed the values where their, I just had to take the trailing .0 off to see them.. Why they are not graphing in Cacti I am still not sure... Might be because I am running it under windows or something.. still looking into it.

Found the problem... The OIDs where missing the period prefix which caused them to fail.

 

The CPU OID is correct, but the flows, memory, TEMP etc are missing it.

 Glad you figured it out. What's interesting is that it has worked that way for most of us, and the original template I made these from also used those values (except on the storage and flow sessions, which I added).  I'll update the template, assuming that adding the dot doesn't break me ;-).  Just kidding, it's obviously a bug in the template that for some reason Cacti has been tolerant of in most situations.

 Would you mind posting your Cacti version? ...actually it probably has more to do with the version of the snmp tools...

I know it's no longer an issue, but my curiosity for understanding details is getting the better of me. If you could post your snmp tools version, that would be helpful to me.  My version doesn't seem to care about whether the prefix is there or not.

 

 

[root@cacti cacti-spine-0.8.7e]# snmpwalk -V
NET-SNMP version: 5.3.2.2

[root@cacti cacti-spine-0.8.7e]# snmpwalk -OS -v2c -c @readonly 10.10.1.1 .1.3.6.1.4.1.2636.3.1.13.1.8.9.1.0.0

SNMPv2-SMI::enterprises.2636.3.1.13.1.8.9.1.0.0 = Gauge32: 13

[root@cacti cacti-spine-0.8.7e]# snmpwalk -OS -v2c -c @readonly 10.10.1.1 1.3.6.1.4.1.2636.3.1.13.1.8.9.1.0.0

SNMPv2-SMI::enterprises.2636.3.1.13.1.8.9.1.0.0 = Gauge32: 12

As I mentioned I am running under windows, in this case for my proof of concept system.. I plan on running under linux when I move to production.

 

 

C:\>snmpwalk -V
NET-SNMP version: 5.5

 

 

I ran the test from command line and got results for both... however in cacti it does not appear to collect or graph any data if that leading period is missing.. I suspect that there is some difference in the windows / linux builds.

 

I am running Cacti 0.8.7e

 

However if you look at the OIDs of all of the built in SNMP graphs they do have the period, and all of the SNMP tools I have used, have required it.

 

Even the EX template you used as a base had the prefixed period for the CPU OID.

 

For compatibility sake it might be good to post an updated template to avoid anyone else having this issue.

 

I do really appreciate the template.. Now if it only included the QoS Queues 😉

Yeah, I've already fixed it. It was wrong, regardless of whether or not it worked.  I was just curious about the differences that caused the inconsistency.

 

 I haven't really used the QoS stuff, and I'm not really sure what portions of it would be useful to the majority of people, otherwise I'd take a stab at it.

 

How where the OIDs determined for the Storage graph? The system health command on the SRX only seems to indicate the exact OIDs for root and config, which really arn't that useful. The current var OID appears to be incorrect for my 650 cluster.

 

Watching all of the OIDs in the storage node also is confusing as the % does not directly match the show system storage command and the order is not the same..

 

We are currently having issues with logs filling the system so any help nailing down the exact OIDs per cluster node would be really useful right now.

 

Also FYI I had to create custom session memory and CPU OIDs for the cluster.. I don't know enough about cacti to write a script that works by index, however it does appear that there is enough information to indicate the name of each node and which OIDs belong to each for the cpu, sessions and memory.. 

The data is found in JUNIPER-HOSTRESOURCES-MIB::jnxHrStorage (.1.3.6.1.4.1.2636.3.31.1), which augments the standard hrStorage MIB (.1.3.6.1.2.1.25.2.3)

Walk the hrStorage and see the hrStorageDescr, they're indexed the same as your jnxHrStoragePercentUsed:

 

 snmpwalk -OS  -v2c -c @readonly 10.10.1.1 .1.3.6.1.2.1.25.2.3
...
HOST-RESOURCES-MIB::hrStorageDescr.1 = STRING: /dev/da0s1a, mounted on: /
HOST-RESOURCES-MIB::hrStorageDescr.2 = STRING: /dev/bo0s3e, mounted on: /config
HOST-RESOURCES-MIB::hrStorageDescr.5 = STRING: devfs: dev file system, mounted on: /dev
HOST-RESOURCES-MIB::hrStorageDescr.6 = STRING: /dev/md0, mounted on: /junos
HOST-RESOURCES-MIB::hrStorageDescr.7 = STRING: /cf, mounted on: /junos/cf
HOST-RESOURCES-MIB::hrStorageDescr.8 = STRING: devfs: dev file system, mounted on: /junos/dev/
HOST-RESOURCES-MIB::hrStorageDescr.9 = STRING: procfs: process file system, mounted on: /proc
HOST-RESOURCES-MIB::hrStorageDescr.10 = STRING: /dev/bo0s3f, mounted on: /cf/var
HOST-RESOURCES-MIB::hrStorageDescr.11 = STRING: /dev/md1, mounted on: /mfs
HOST-RESOURCES-MIB::hrStorageDescr.12 = STRING: /cf/var/jail, mounted on: /jail/var
HOST-RESOURCES-MIB::hrStorageDescr.13 = STRING: /cf/var/log, mounted on: /jail/var/log
HOST-RESOURCES-MIB::hrStorageDescr.14 = STRING: devfs: dev file system, mounted on: /jail/dev
HOST-RESOURCES-MIB::hrStorageDescr.15 = STRING: /dev/md2, mounted on: /mfs/var/run/utm
HOST-RESOURCES-MIB::hrStorageDescr.16 = STRING: /dev/md3, mounted on: /jail/mfs
...

  match indexes with:

snmpwalk -OS  -v2c -c @readonly 10.10.1.1 .1.3.6.1.4.1.2636.3.31.1
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.1 = Gauge32: 64
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.2 = Gauge32: 0
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.5 = Gauge32: 100
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.6 = Gauge32: 100
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.7 = Gauge32: 64
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.8 = Gauge32: 100
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.9 = Gauge32: 100
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.10 = Gauge32: 57
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.11 = Gauge32: 8
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.12 = Gauge32: 57
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.13 = Gauge32: 57
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.14 = Gauge32: 100
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.15 = Gauge32: 0
SNMPv2-SMI::enterprises.2636.3.31.1.1.1.1.16 = Gauge32: 0

 

 

For some reason the numbers tend to be off by a few percent if you compare to a 'show system storage', not sure if that's a rounding error or what, but it's close enough to keep an eye on when things are running out.

Prefect.. FYI the /var partitioning appears to only differ if your device has the older style single boot option vs the new 10.x dual boot.. They seem to have moved a LOT MORE SPACE to the /var partition which might actually help us with our logging problems.

FYI

 

Template downloaded today on a SRX240 with JunOS 10.3R1.9 on cacti 0.8.7g gets flows and storage, but no CPU, temp or memory.

 

Has any updates been done recently to this template? It's working great with the exception of session flows and temperature.  I just recently swapped out our ASA with an SRX3400 and I am looking for some tools to monitor usage etc.

 

Thanks in advance.

Hey buddy,

 

Your template is really helpful. I got graphs for SRX650 but for SRX14oo i am not able to get the graphs. I think maybe the OID's or MIB's are diff. How and where do i edit the SNMP OID's for 1400 in your template as i am not sure on that part. Any help will be appreciated.

 

Arien

You can change it directly in cacti under Data Templates   look for the Juniper SRX templates

 

[edit: Juniper SRX - Memory Utilization]

 

Look for the OID section and you will see the "default used OID" which you can change

Hey buddy,

 

Will this break the other SRX chassis i.e 240, 650 etc as the template is common for all SRX. OR are you doing separate templates for all the SRX.

 

Raghav

You can copy the Data Template and rename it to data_template_srx650 and change the oid then! You can test them on both the 650 and 240 to see if it works.

Hi,

 

First, thank you very much for this template, it is awesome.

 

But i have a question about the flo graph, it is not working for me.

 

I have incoming flows and can get statistics with the FlowView Plugin from CactiEZ, but the graph does not count.

 

Can someone help me to get this graph working? Is there addional configuration required?

 

best regards from austria

wamon