ExtremeCloud IQ- Site Engine & Extreme Management Center

Expand all | Collapse all

Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

Stephane Grosjean

Stephane Grosjean07-03-2018 10:38

  • 1.  Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 07:24
    Hello everybody

    I would like to configure the Extreme Switches (x440-G2, with version 22.4) to sending sflow/netflow data direct to the Extreme Analytics. I heard this is now possible without having a flowcollector enginge like PV FC-180 installed. Is this true? Because I couldn't find any referenced documentation to this. Thank you very much for your feedback.

    Best regards, Yves


  • 2.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 07:34
    Hello, I have made this working in my lab using Analytics 8.1.1EA2 and it is possible,. Some manual actions are needed on XMC but after that it can work. You could run into problems due to ACL limitations on the X440G2 but it can work.


  • 3.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 07:38
    Hi Oskar, I have the same Analytics version running. Do you have a configuration example or a documentation of this? Thank you.



  • 4.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 08:04
    Ok, some file editing is needed now to make it work.
    On XMC, copy sflow,pol.
    cp /usr/local/Extreme_Networks/NetSight/appdata/Purview/Fingerprints/sflow.pol /tftpboot/

    On Analytics, edit file /opt/appid/conf/appidconfig.xml and change interfaces to be like:
    [i]

    [i]



    Restart analytics or appid after this.

    On XMC go to analytics, configuration and on your engine add an application telemetry source (the switch) and enforce.



  • 5.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 08:04
    Hi Oskar, it's me again. Are we talking here about sflow+ ? Because my workmate said, it has to run with sflow+ and not only with sflow.


  • 6.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 08:04
    sflow together with a mirror config.

    TestAnalytics.1 # sh sflow
    SFLOW Global Configuration
    Global Status: enabled
    Polling interval: 60
    Sampling rate: 1024
    Maximum cpu sample limit: 2000
    SFLOW Configured Agent IP: 10.116.3.89 Operational Agent IP: 10.116.3.89
    Collectors
    Collector IP 10.116.2.209, Port 6343, VR "VR-Default"

    SFLOW Port Configuration
    Port Status Sample-rate Subsampling Sflow-type
    Config / Actual factor Ingress / Egress
    1 enabled 1024 / 1024 1 enabled / disabled
    2 enabled 1024 / 1024 1 enabled / disabled
    TestAnalytics.2 # sh mirror

    DefaultMirror (Disabled)
    Description: Default Mirror Instance, created automatically
    Mirror to port: -

    EAN (Enabled)
    Description:
    Mirror to remote IP: 10.116.2.209 VR : VR-Default
    From IP : 10.116.3.89 Ping check: Off
    Status : Up

    Mirrors defined: 2
    Mirrors enabled: 1 (Maximum 4)
    HW filter instances used: 0 (Maximum 128)



  • 7.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 08:14
    I don't have a sflow.pol file.
    And how/where can I add a "application telemetry source"?

    What do I have to configure on the switch side?

    Thank you


  • 8.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 08:17
    You add the app telemetry source under flow sources.
    Maybe the sflow.pol file is only created once you do a first enforce.
    On the switch I think all should be configured by Analytics.


  • 9.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 02-28-2018 08:23
    Okay, found it, thx.
    I did an enforce (without adding a telemetry source) but the file isn't there yet.

    I guess I have to test it first in the lab, before I add a switch to the Analytics on customer site. Because I have no idea how business critical this is.


  • 10.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 03-02-2018 13:55
    I tried it with "Application Telemetry Source" (I guess this is for SFlow+).
    But I got the following errors when I add the source Switch.

    -> if {! $OverallResult} {
    2018-03-02 15:51:52,183 INFO [stdout] Script failed : configure access-list sflow.pol any ingress
    2018-03-02 15:51:52,185 INFO [stdout] .
    2018-03-02 15:51:52,185 INFO [stdout] Error: ACL install operation failed - filter hardware full for vlan *, port *
    2018-03-02 15:51:52,185 INFO [stdout] * switchname.14 #
    2018-03-02 15:51:52,188 ERROR [com.enterasys.netsight.appid.server.webapps.monitor.AppIdDwr] Error in step 2 of enabling SFlow+ source for 10.37.1.52
    2018-03-02 15:51:52,192 ERROR [com.enterasys.netsight.appid.server.webapps.monitor.AppIdDwr] javax.script.ScriptException:
    *** Script Error ***
    Die command issued: Script failed : configure access-list sflow.pol any ingress
    .
    Error: ACL install operation failed - filter hardware full for vlan *, port *
    * switchname.14 #

    --> I've already run this command: "configure access-list vlan-acl-precedence shared" and rebooted the switch.

    The configuration is a default setup.



  • 11.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 03-02-2018 15:39
    Have the same problem on X440-G2:
    Error: ACL install operation failed - filter hardware full for vlan *, port *


  • 12.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 03-02-2018 15:39
    this is a known CR, that will be fixed (if not already)


  • 13.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 03-12-2018 07:07
    Try to configure the access-list width to double.
    The command to do this is:

    configure access-list width double



  • 14.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 04-06-2018 08:02
    Hi Oscar,

    I've also test this, but gets the same error as Yves & Bernhard. I've used 22.4.1.4-patch1-2, but we're also working with Policy enabled. So couldn't run the command:

    configure access-list width double slot 1

    WARNING: Configuration of double width access-list is not supported when Policy is enabled. Slots will remain in Single width mode.


  • 15.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 04-06-2018 09:22
    Hello everybody: I got the following information: To use an ExtremeSwitching X440-G2 switch as an Application Telemetry source for ExtremeAnalytics, install firmware version 22.4.1.4-patch2-5 or higher.


  • 16.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 04-06-2018 10:13
    Hi Yves,

    Ok thanks for the information. The firmware 22.4.1.4-patch2-5 isn't yet available at the partner portal of Extreme Networks?

    Kind Regards,
    Kevin.


  • 17.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 04-24-2018 06:20
    Hello Kevin, 22.4.1.4-patch2-5 is a private patch and is not publically available. You can get this patch by opening a case.
    22.5 will also contain the fix once it is available (expected end of May).


  • 18.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 04-24-2018 06:24
    Hi Oscar,

    Ok thanks for the information, I will contact Extreme GTAC.

    Kind Regards,
    Kevin.


  • 19.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    Hi,

    Just to let you know I'm running into this same problem using firmware version:

    22.5.1.7

    EMC = 8.1.3.65

    Error is:

    An error has occurred while adding a flow source to the selected engine. See details below for more information.

    javax.script.ScriptException: *** Script Error *** Die command issued: Script failed : * POC.17 # configure access-list telemetry.pol any ingress . Error: ACL install operation failed - filter hardware full for vlan *, port * * POC.18 # -> if {! $OverallResult}

    Am also running policy on this X440G2 switch, so setting the double width I get the following message:

    * POC.1 # configure access-list width double
    WARNING: Configuration of double width access-list is not supported when Policy is enabled. Slots will remain in Single width mode.

    If I disable policy I can then run the access-list width double command, but when I re-enable policy I then get the following error:

    * POC.4 # enable policy
    Warning: Enabling Policy will cause some Netlogin settings (such as VLANs and dynamically created VLANs) to be cleared.
    ERROR: Hardware resources could not be reserved for Policy (count 0).
    Note that Policy cannot be enabled when double width access-list is configured or operational.

    Previously upgraded from version 22.4.1.4-patch1-2

    Post mentions this being fixed in version 22.5, so not sure what I am doing wrong or how to fix / get this working?

    Seen this command that could be used to adjust ACL usage, but not sure exactly the syntax to use that would help:

    configure policy resource-profile default profile-modifier [no-mac|no-ipv4|no-ipv6] enable

    I've tried removing no-ipv6, but no difference?

    Many thanks.


  • 20.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    Hi,
    can you show the output of "show sflow hardware-utilization"?


  • 21.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    Hi Stephane,

    See output below:

    POC.4 # show sflow hardware-utilization
    sFlow Hardware Table Utilization Statistics
    Resource Type Current Maximum % Util.
    -------------------- ------- ------- -------
    MAC Entries 11 16383 0
    Host Entries 0 4096 0
    IPv4 Entries 0 4096 0
    IPv6 Entries 0 2048 0
    Long IPv6 Entries 0 2048 0
    Total Routes 16 512 3
    IPv4 Neighbors 5 N/A N/A
    IPv6 Neighbors 0 N/A N/A
    IPv4 Routes 16 N/A N/A
    IPv6 Routes 0 N/A N/A
    ECMP Next Hops 0 N/A N/A
    ACL Ingress Entries 86 2048 4
    ACL Ingress Counters 34 1024 3
    ACL Ingress Meters 0 2048 0
    ACL Ingress Slices 7 8 87
    ACL Egress Entries 0 512 0
    ACL Egress Counters 0 512 0
    ACL Egress Meters 0 512 0
    ACL Egress Slices 0 4 0

    Thanks


  • 22.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    you use 87% of the ACL Ingress Slices, so you may be using too much resources. You could see more info with that command:

    sh access-list usage acl-slice port 1:1


  • 23.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    Hi Stephane,

    Thanks for posting back. I've currently disabled policy and re-run the Telemetry configuration via EMC, just to see it would take, which it did.

    Problem is I can't enable Policy again because its out of resource. So least I know it works with one or the other but never both.

    Here is were it stands at the moment with sflow telemetry enabled and policy disabled:

    POC.2 # sh access-list usage acl-slice port 1
    Ports 1-24, 51, 52
    Stage: INGRESS
    Slices: Used: 4 Available: 4
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 256
    Virtual Slice 4 (physical slice 4) Rules: Used: 12 Available: 244 system
    Virtual Slice 5 (physical slice 5) Rules: Used: 2 Available: 254 system
    Virtual Slice 6 (physical slice 6) Rules: Used: 123 Available: 133 user/other
    Virtual Slice 7 (physical slice 7) Rules: Used: 71 Available: 185 user/other
    Stage: EGRESS
    Slices: Used: 1 Available: 3
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128
    Virtual Slice 3 (physical slice 3) Rules: Used: 1 Available: 127 user/other
    Stage: LOOKUP
    Slices: Used: 0 Available: 4
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128
    Stage: EXTERNAL
    Virtual Slice : (*) Physical slice not allocated to any virtual slice.

    If I disable the ACL's for telemetry (show config acl)

    configure meter ingmeter2 committed-rate 1024 Kbps out-actions drop ports "Default (IRL.1)"
    configure meter ingmeter3 committed-rate 2048 Kbps out-actions drop ports "Default (IRL.1)"
    configure access-list telemetry any ingress
    configure access-list telemetryegress any egress

    i.e

    unconfigure access-list telemetry
    unconfigure access-list telemetryegress

    and re-enable policy the ACL slices look like the following:

    POC.7 # sh access-list usage acl-slice port 1
    Ports 1-24, 51, 52
    Stage: INGRESS
    Reserved slices:
    Type Used Available
    Policy P/D 0 2
    Policy CoS 0 2
    Slices: Used: 7 Available: 1
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 256 Policy CoS reserved
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 256 Policy CoS reserved
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 256 Policy P/D reserved
    Virtual Slice * (physical slice 6) Rules: Used: 0 Available: 256 Policy P/D reserved
    Virtual Slice 5 (physical slice 4) Rules: Used: 12 Available: 244 system
    Virtual Slice 6 (physical slice 5) Rules: Used: 2 Available: 254 system
    Virtual Slice 7 (physical slice 7) Rules: Used: 64 Available: 192 user/other
    Stage: EGRESS
    Slices: Used: 0 Available: 4
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128
    Stage: LOOKUP
    Reserved slices:
    Type Used Available
    Policy Profile 0 4
    Slices: Used: 4 Available: 0
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128 Policy Profile reserved
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128 Policy Profile reserved
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128 Policy Profile reserved
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128 Policy Profile reserved
    Stage: EXTERNAL
    Virtual Slice : (*) Physical slice not allocated to any virtual slice.

    With both policy and telemetry disabled:

    * POC.9 # sh access-list usage acl-slice port 1
    Ports 1-24, 51, 52
    Stage: INGRESS
    Slices: Used: 3 Available: 5
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 256
    Virtual Slice * (physical slice 6) Rules: Used: 0 Available: 256
    Virtual Slice 5 (physical slice 4) Rules: Used: 12 Available: 244 system
    Virtual Slice 6 (physical slice 5) Rules: Used: 2 Available: 254 system
    Virtual Slice 7 (physical slice 7) Rules: Used: 64 Available: 192 user/other
    Stage: EGRESS
    Slices: Used: 0 Available: 4
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128
    Stage: LOOKUP
    Slices: Used: 0 Available: 4
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128
    Stage: EXTERNAL
    Virtual Slice : (*) Physical slice not allocated to any virtual slice.

    Is there anyway to see what's using those 3 slices when both policy and telemetry are disabled, maybe those meters?



  • 24.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    for testing purposes, I'd try to disable the meters, and try to use a light Policy with Telemetry. If possible, can you make sure for your Policy TCI Overwrite is disabled as well as CoS (if not needed, of course)?


  • 25.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    There are no TCi Overwrites configured and removed all the CoS settings, as can been seen in the screenshot below for one example I set the CoS to 'none'



    But even after an enforce the cos value stays persistent in the policy configuration on the switch?



    So manually removed this but still no difference to amount of ACL slices used.


  • 26.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    You should not see that line imho.


  • 27.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    No worries. Thanks for helping out on this Stephane. It is quite a challenge, just doesn't seem a clear way to see what's using the slices. Even when policy is disabled and all the ACL config is removed 3 are still consumed, but not sure what?

    Tried everything I can fined from compression, shared, dedicated, double ACLs, but nothing is effecting or reducing the values.

    There is definitely a dark art to it somewhere :)


  • 28.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    fwiw, you should be able to have telemetry and policy on the x440G2, but I doubt you can have CoS settings as well. You should check why you still have a cos configured. did you try without the meters?


  • 29.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    Well..

    I've disable CoS in policy and did an enforce, well I think that's what I've done - see below:



    When you look on the switch there is a lot of config that policy originally added, and it doesn't get removed, even removing a CoS from a role (had to do that manually as previously stated):

    configure ports group "Default (IRL.1)" add 1-52
    configure ports group "Default (TXQ.0)" add 1-52
    configure qosscheduler strict-priority ports "Default (TXQ.0)"
    configure qosprofile QP1 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure qosprofile QP2 maxbuffer 100 weight 1
    configure qosprofile QP2 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure qosprofile QP3 maxbuffer 100 weight 1
    configure qosprofile QP3 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure qosprofile QP4 maxbuffer 100 weight 1
    configure qosprofile QP4 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure qosprofile QP5 maxbuffer 100 weight 1
    configure qosprofile QP5 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure qosprofile QP6 maxbuffer 100 weight 1
    configure qosprofile QP6 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure qosprofile QP7 maxbuffer 100 weight 1
    configure qosprofile QP7 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure qosprofile QP8 maxbuffer 100 weight 1 ports "Default (TXQ.0)"
    configure dot1p type 0 qosprofile QP1 ingress-meter ingmeter0
    configure dot1p type 1 qosprofile QP2 ingress-meter ingmeter1
    configure dot1p type 2 qosprofile QP3 ingress-meter ingmeter2
    configure dot1p type 3 qosprofile QP4 ingress-meter ingmeter3
    configure dot1p type 4 qosprofile QP5 ingress-meter ingmeter4
    configure dot1p type 5 qosprofile QP6 ingress-meter ingmeter5
    configure dot1p type 6 qosprofile QP7 ingress-meter ingmeter6
    configure dot1p type 7 qosprofile QP8 ingress-meter ingmeter7
    configure vr VR-Default delete protocol mpls
    configure cos-index 8 qosprofile QP4 replace-tos 64
    configure meter ingmeter2 committed-rate 10000000 Kbps out-actions drop ports "Default (IRL.1)"
    configure meter ingmeter3 committed-rate 10000000 Kbps out-actions drop ports "Default (IRL.1)"
    configure access-list vlan-acl-precedence shared
    configure access-list rule-compression port-counters shared

    So apologies if this might seem like a daft question but do you know how I can fully remove all the CoS settings, perhaps just by manually removing all the config above?

    I tried removing the meters but it complained about being in use elsewhere, probably the 'configure dot1p' command.

    Before I start removing all the config, and its only a POC so its no big deal, I just wanted to see if there was an easier way and that I'm going about it correctly.

    Many thanks.



  • 30.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    A fresh x440G2 (default config) should use only 1 slice.

    X440G2-24p-10G4.1 # show access-list usage acl-slice port 1

    Ports 1-28

    Stage: INGRESS

    Slices: Used: 1 Available: 7

    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 256

    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 256

    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 256

    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 256

    Virtual Slice * (physical slice 4) Rules: Used: 0 Available: 256

    Virtual Slice * (physical slice 5) Rules: Used: 0 Available: 256

    Virtual Slice * (physical slice 6) Rules: Used: 0 Available: 256

    Virtual Slice 7 (physical slice 7) Rules: Used: 10 Available: 246 system

    Stage: EGRESS

    Slices: Used: 0 Available: 4

    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128

    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128

    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128

    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128

    Stage: LOOKUP

    Slices: Used: 0 Available: 4

    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128

    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128

    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128

    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128

    Stage: EXTERNAL



    Telemetry + Policy should leave you with 1 slice free.

    You could try disabling diffserv examination and dot1p examination and turn off port qos (config port qosprofile none). If all else fails, you can reduce policy to just L2 rules via:



    config policy resource-profile default profile-modifier no-ipv4 enable



    This will require disable/enable policy but will effectively reserve/consume only 2 slices (instead of 4) for policy.



  • 31.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    btw, you are not using double width ACL, right? You should not.


  • 32.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    Thanks Stephane, shall try this out and post back. I haven't got double width enabled. I did try it, but removed it as I couldn't enable policy.


  • 33.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    So I've tried all the above and removed all the QoS config I can, but still the ACL slices will not budge, so no idea what's using them?

    Think the only way I'm going to be able to straighten it out I think is wipe the switch and start again.

    Thanks for all your effort anyway.


  • 34.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 10:38
    that's certainly a wise decision, playing all around with the ACL config as you did, you may have some leftovers.

    Make sure CoS is disabled in Policy Manager, like that:





  • 35.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 07-03-2018 18:16
    For what its worth I tried added these commands to minimize use of the ACL slices:

    configure access-list vlan-acl-precedence shared
    configure access-list rule-compression port-counters shared

    But made no difference?



  • 36.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 11-21-2018 12:37
    Has this been fixed? I just tried to add my x440G2 into Analytics into a lab environment where I'm also running control doing some basic mac-based authentication at the port...and I get the same error when I try and add the 440 to the analytics engine:

    2018-11-21 11:32:37,434 ERROR [com.enterasys.netsight.appid.server.webapps.monitor.AppIdDwr] javax.script.ScriptException:
    *** Script Error ***
    Die command issued: Script failed : * X440G2-12p-10G4.17 # configure access-list telemetry.pol any ingress

    Error: ACL install operation failed - filter hardware full for vlan *, port *
    * X440G2-12p-10G4.18 #


  • 37.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 11-21-2018 13:33
    Did you tried this command: configure access-list vlan-acl-precedence shared
    --> and reboot.


  • 38.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 11-23-2018 10:48
    Just tried it....I get the same failure, posted below. My SE said this should work fine...as well as with two VSP 4450s I have in my lab environment. I can see flows, but only by typing in the sflow instructions into the CLI of the VSPs and the 440G2...I cannot seem to have analytics automatically configure them via XMC because it doesn't even see the VSPs, and the error when I try and add it to the app telemetry section of the engine:

    2018-11-23 12:45:23,159 ERROR [com.enterasys.netsight.appid.server.webapps.monitor.AppIdDwr] javax.script.ScriptException:
    *** Script Error ***
    Die command issued: Script failed : * X440G2-12p-10G4.17 # configure access-list telemetry.pol any ingress

    Error: ACL install operation failed - filter hardware full for vlan *, port *
    * X440G2-12p-10G4.18 #



  • 39.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 05-04-2019 23:10
    Same here with 22.6.1 patch 1.1.
    Couple of X440-G2 stacks, all are ok but one.
    Policy (simple at the moment, not much rules, HTTP redirect, no qos yet qos enabled on each), telemetry.
    Telemetry was ok on all of these but on one I was receiving a lot of such logs, for different ACL entries:
    code:
    HAL.IPv4ACL: Slot-1: Mirroring instance EAN was disabled, hence mirror action in telemetry policy entry CITRIXICA was disabled on port 1:1


    I've decided to re-add this telemetry source but was unsuccessful, not sure if because of an error in XMC like this (based on my browser history):
    code:
    EXOS application attempting to install incompatible ACL: filter vlan *, port * (rule "bjnp", index 14)


    Rebooted the switch (btw after trying to add a telemetry source before the stack got config internally synced, it disappeared from a list of allowed devices, I had to reboot entire XMC eventually) and... ah (...), here we go again. ;)
    No luck with changing vlan-acl-precedence, no luck with removing qos meters and disabling/enabling CoS via Policy in XMC.

    Drop some CLI:
    code:
    Slot-1 SW_STACK-1.45 # show access-list usage acl-slice port 1:1
    Ports 1:1-1:24, 1:51, 1:52
    Stage: INGRESS
    Reserved slices:
    Type Used Available
    Policy P/D 1 1
    Policy CoS 0 2

    Slices: Used: 7 Available: 1
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 256 Policy CoS reserved
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 256 Policy CoS reserved
    Virtual Slice * (physical slice 5) Rules: Used: 0 Available: 256 Policy P/D reserved
    Virtual Slice * (physical slice 6) Rules: Used: 0 Available: 256
    Virtual Slice 4 (physical slice 0) Rules: Used: 14 Available: 242 system
    Virtual Slice 5 (physical slice 1) Rules: Used: 2 Available: 254 system
    Virtual Slice 6 (physical slice 4) Rules: Used: 76 Available: 180 Policy P/D reserved
    Virtual Slice 7 (physical slice 7) Rules: Used: 4 Available: 252 user/other
    Stage: EGRESS
    Slices: Used: 0 Available: 4
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128
    Stage: LOOKUP
    Reserved slices:
    Type Used Available
    Policy Profile 0 4

    Slices: Used: 4 Available: 0
    Virtual Slice * (physical slice 0) Rules: Used: 0 Available: 128 Policy Profile reserved
    Virtual Slice * (physical slice 1) Rules: Used: 0 Available: 128 Policy Profile reserved
    Virtual Slice * (physical slice 2) Rules: Used: 0 Available: 128 Policy Profile reserved
    Virtual Slice * (physical slice 3) Rules: Used: 0 Available: 128 Policy Profile reserved
    Stage: EXTERNAL

    Virtual Slice : (*) Physical slice not allocated to any virtual slice.

    * Slot-1 SW_STACK-1.46 # show access-list dynamic
    Dynamic Rules: ((*)- Rule is non-permanent )

    (*)hclag_arp_2_4_96_9e_e3_41 Bound to 0 interfaces for application HealthCheckLAG
    (*)policy.rule.df.1.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.10 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.10.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.11.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.12.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.13 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.13.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.14.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.15.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.16.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.17.22.8 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.18.22.8 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.19.25.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.2.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.20.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.21.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.22.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.23.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.24.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.25.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.26.25.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.27.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.28.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.29.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.3.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.30.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.31.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.32.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.33.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.34.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.35.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.36.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.37.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.38.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.39.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.4 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.4.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.40.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.41.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.42.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.43.22.8 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.44.22.8 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.45.25.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.46.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.47.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.48.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.49.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.5 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.5.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.50.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.51.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.52.25.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.53.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.54.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.55.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.56.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.57.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.58.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.59.25.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.6 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.6.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.60.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.61.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.62.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.63.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.64.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.65.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.66.25.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.67.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.68.16.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.69.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.7.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.70.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.71.25.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.8.18.16 Bound to 1 interfaces for application Policy
    (*)policy.rule.df.9.18.16 Bound to 1 interfaces for application Policy
    (*)policy.webredir.server.13.0 Bound to 1 interfaces for application Policy
    (*)policy.webredir.sock.13.0 Bound to 1 interfaces for application Policy
    (*)policy.webredir.sock.13.1 Bound to 1 interfaces for application Policy
    (*)policy.webredir.sock.13.2 Bound to 1 interfaces for application Policy
    * Slot-1 SW_STACK-1.47 # show sflow hardware-utilization

    sFlow Hardware Table Utilization Statistics

    Slot: 1
    Type: X440G2-48p-10G4
    Resource Type Current Maximum % Util.
    -------------------- ------- ------- -------
    MAC Entries 387 16384 2
    Host Entries 16 4096 0
    IPv4 Entries 0 4096 0
    IPv6 Entries 0 2048 0
    Long IPv6 Entries 0 2048 0
    Total Routes 2 512 0
    IPv4 Neighbors 1 N/A N/A
    IPv6 Neighbors 0 N/A N/A
    IPv4 Routes 2 N/A N/A
    IPv6 Routes 0 N/A N/A
    ECMP Next Hops 0 N/A N/A
    ACL Ingress Entries 96 2048 4
    ACL Ingress Counters 0 2048 0
    ACL Ingress Meters 0 2048 0
    ACL Ingress Slices 7 8 87
    ACL Egress Entries 0 512 0
    ACL Egress Counters 0 512 0
    ACL Egress Meters 0 512 0
    ACL Egress Slices 0 4 0

    Slot: 2
    Type: X440G2-48t-10G4
    Resource Type Current Maximum % Util.
    -------------------- ------- ------- -------
    MAC Entries 388 16384 2
    Host Entries 16 4096 0
    IPv4 Entries 0 4096 0
    IPv6 Entries 0 2048 0
    Long IPv6 Entries 0 2048 0
    Total Routes 0 512 0
    IPv4 Neighbors 0 N/A N/A
    IPv6 Neighbors 0 N/A N/A
    IPv4 Routes 0 N/A N/A
    IPv6 Routes 0 N/A N/A
    ECMP Next Hops 0 N/A N/A
    ACL Ingress Entries 95 2048 4
    ACL Ingress Counters 0 2048 0
    ACL Ingress Meters 0 2048 0
    ACL Ingress Slices 7 8 87
    ACL Egress Entries 0 512 0
    ACL Egress Counters 0 512 0
    ACL Egress Meters 0 512 0
    ACL Egress Slices 0 4 0

    ... up to the rest of slots


    I've lost any clue for the moment based on the fact that couple of stacks have the same EXOS version and Policy domain and there are no issues with Telemetry. :(

    Did somebody get through it recently?

    Kind regards,
    Tomasz


  • 40.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 05-04-2019 23:19
    [it doesn't let me edit the post] I see two slices reserved for CoS but I hope there is a workaround different than default&redo...


  • 41.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 05-09-2019 22:56
    Hi ,

    Were you successful with the setup over that time? I was successful now thanks to our local SE advice, here's what I did to get this working on the last stack of X440-G2s:
    • uploaded configuration as script,
    • left essentials, without any Policy, AAA and named the file default.xsf,
    • put the script back to the switch and did unconfigure switch; the script started just a moment after booting from defaults so I had minimum downtime,
    • set the stack as a telemetry source in Analytics first,
    • removed the stack from Policy domain (otherwise had issues), saved, added back, saved, enforced,
    • re-enforced NAC appliance with "force reconfiguration for all switches" just to be sure I didn't remove too much of it when preparing default.xsf parachute.
    That helped, but still gotta have in mind that X440-G2 has quite limited resources for all-inclusive scenario.

    The only thing that concerns me now is 75 pages of syslog in XMC with that kind of messages just for this 1 out of 8 stacks:
    code:
    HAL.IPv4ACL: Slot-1: Mirroring instance EAN was disabled, hence mirror action in telemetry policy entry BITTORRENT was disabled on port 1:1



    I thought it stopped but after a few moments does the same for tons of entries and for different ports.
    Show mirror when it happens:
    code:
    DefaultMirror   (Disabled)
    Description: Default Mirror Instance, created automatically
    Mirror to port: -

    EAN (Enabled)
    Description:
    Mirror to remote IP: x.x.x.x VR : VR-Default
    From IP : y.y.y.y Ping check: On
    Status : Up

    Mirrors defined: 2
    Mirrors enabled: 1 (Maximum 4)
    HW filter instances used: 0 (Maximum 128)
    HW mirror instances used: 0 ingress, 0 egress (Maximum 4 total, 1 egress)



    Will keep an eye on this but if somebody already walked through this that would be much appreciated. :)

    Kind regards,
    Tomasz


  • 42.  RE: Sending flow data from a switch direct to ExtremeAnalytics (without a flowcollector engine)

    Posted 05-22-2019 11:15
    Hi,

    FYI it was an ultimate example why one should not perform full-blown config for X440-G2s in a stack that exceeds best practice size. The stack that made issues was 7-slot, others were 2-4 slots only. Because of that most likely, the master's CPU was heavily loaded, reaching 100% quite often. Probably that's why we've seen many logs about mirroring instances, and also ports started to flap (at least those with APs connected). Huge mess. ;)
    Disabling telemetry left the switch free of high CPU utilization and AP link outages, next step is to simply split the stack.

    Hope that helps,
    Tomasz