This PR is cherry-pick of master
https://github.com/Azure/sonic-buildimage/pull/6920
Why I did it
Add support for BGP Monitors on multi asic SONiC platforms.
How I did it
On multi ASIC SONiC platforms, BGP monitor session will be established from Backend ASIC.
To achieve this following changes are done
Add BGP monitor configuration on the backend ASIC.
The BGP monitor configuration is present in the DPG of the device in minigraph.xml of multi-ASIC device, so this configuration will be added to the config_db of the host, when the minigraph is loaded.
To add configuration for this in the Backend ASIC, a new class MultiAsicBgpMonCfg is added to the hostcfgd service to update the config_db of the backend ASIC when the BGP_MONITOR table of the host config_db is updated.
This way incremental BGP_MONITOR configuration can also be handled.
Changes to establish BGP session with bgp monitor.
Add route in host main routing table to go to one of pre-define backend asic
Add IP table rule on front asic to mark the BGP packets with destination as IPv4 Loopback.
Add IP rule in front asic namespace to match mark BGP packet and lookup default table
Program the default route in FrontEnd asic name space docker default table as part of start.sh of the BGP container.
It need to be done as part of start.sh otherwise FRR default route will get over-written.
How to verify it
Signed-off-by: Abhishek Dosi <abdosi@microsoft.com>
Co-authored-by: Arvind <arlakshm@microsoft.com>
To run VNET route consistency check periodically.
For any failure, the monit will raise alert based on return code.
The tool will log required details.
- [201911][acl] Expand VLAN into VLAN members when creating an ACL table (#1477)
- [201911][acl-loader] Add support for matching on ICMP and VLAN info (#1476)
- [201911][acl-loader] Improve input validation for acl_loader (#1481)
Signed-off-by: Danny Allen <daall@microsoft.com>
#### Why I did it
Change buffer config for new SKU Mellanox-SN2700-D40C8S8
#### How I did it
Reuse the buffer config of SKU Mellanox-SN2700-D48C8
#### How to verify it
Run sonic-mgmt qos test and all passed
This is to address the issue when "mmuconfig -p egress_lossy_profile" is executed which causes SYNCd failure with SAI_STATUS_INSUFFICIENT_RESOURCE for attr SAI_BUFFER_PROFILE_ATTR_RESERVED_BUFFER_SIZE.
This change also requires the change from (https://github.com/Azure/sonic-swss/pull/1649)
This SAI change was already tested as part of the (https://github.com/Azure/sonic-swss/pull/1649) PR.
1. Made the command next-hop-self force only applicable on back-end asic bgp. This is done so that BGPL iBGP session running on backend can send e-BGP learn nexthop. Back end asic FRR is able to recursively resolve the eBGP nexthop in its routing table since it knows about all the connected routes advertise from front end asic.
2. Made all front-end asic bgp use global loopback ip (Loopback0) as router id and back end asic bgp use Loopbacl4096 as ruter-id and originator id for Route-Reflector. This is done so that routes learnt by external peer do not see Loopback4096 as router id in show ip bgp <route-prerfix> output.
3. To handle above change need to pass Loopback4096 from BGP manager for jinja2 template generation. This was missing and this change/fix is needed for this also https://github.com/Azure/sonic-buildimage/blob/master/dockers/docker-fpm-frr/frr/bgpd/templates/dynamic/instance.conf.j2#L27
4. Enhancement to add mult_asic specific bgpd template generation unit test cases.
Enable BBR config allowas-in 1 for internal peers
Why I did:
To advertise BBR routes learnt via e-BGP peer in one asic/namespace to another iBGP asic/namespace via Route Reflector.
Install pyroute2 need for sonic-utilities in sonic-slave-stretch docker.
Submodule update of sonic-utilities to the commit 9297d5c5a00e64b5dea94a49a69cb776ac862bdc
The Portchannels were not getting cleaned up as the cleanup activity was taking more than 10 secs which is default docker timeout after which a SIGKILL will be send.
Fix Issue #6537
- Why I did it
Current mutli-asic vs hwsku consists of 6 asics with each asic having 32 interfaces. When bringing this up, below issue was seen:
When all 32 interfaces(sonic interfaces and linux interface) are set to 9100 mtu, DMA error is seen "DMA: Out of SW-IOMMU space for 4096 bytes at device 0000:06:03.0" which can be fixed by updating swiotlb=65536 in /host/grub/grub.cfg .In order to keep multi-asic VS lighter and easier to bring up and test, new hwsku 'msft_four_asic_vs' is added to represent 4-asic hwsku with 2 frontend asics and 2 backend asics and each asic having 8 interfaces interconnected by port-channels.
- How I did it
Add msft_four_asic_hwsku directory to have the right number of directories (4) and update port_config.ini and lanemap.ini files to include 8 ports information.
Add topology.sh script to create the internal asic-asic connectivity.
- How to verify it
Update asic.conf with the 4 asic information as below and build sonic-vs.img:
NUM_ASIC=4
DEV_ID_ASIC_0=0
DEV_ID_ASIC_1=1
DEV_ID_ASIC_2=2
DEV_ID_ASIC_3=3
Modify sonic_multiasic.xml to have 8 front panel interfaces.
create virtual switch using "sudo virsh sonic_mutliasic.xml" command.
Start topology service and Load config_db files for switch and each asic.
Ensure that that all internal interfaces and port_channels are coming up.
multi-asic vs testbed:
Bring up mutli-asic VS testbed with a multi-asic image(asic.conf updated to 4 asics) and using t1-lag topology.
./testbed-cli.sh -t vtestbed.csv -m veos_vtb -k ceos add-topo vms-kvm-four-asic-t1-lag password.txt
Load minigraph/config_dbs.
Ensure all internal and external interfaces come up.
No change on single asic vs.
61acd3a2e4a457f3bc706cbfaf3162b947763864 (HEAD -> 201911, origin/201911)
[xcvrd] Change in xcvrd ports cache creation, now ports are being
fetched from config DB (#5892) (#155)
Signed-off-by: Abhishek Dosi <abdosi@microsoft.com>
Current mutli-asic vs hwsku consists of 6 asics with each asic having 32 interfaces.
When bringing this up, below issue was seen:
When all 32 interfaces in each namespace (sonic interfaces and linux interface) is set to 9100 mtu, DMA error is seen "DMA: Out of SW-IOMMU space for 4096 bytes at device 0000:06:03.0" which can be fixed by updating swiotlb=65536 in /host/grub/grub.cfg .
Signed-off-by: SuvarnaMeenakshi <sumeenak@microsoft.com>
Update topology script to retrieve hwsku from minigraph
if hwsku information is not available in config_db.
Fix clean up of interfaces in msft_multi_asic_vs hwsku
topology script.
- Why I did it
When bringing up multi-asic VS switch, topology service is started during boot up.
Topology service starts a shell script which runs the topology script present in /usr/share/sonic/device// directory. To invoke hwsku specific script, the topology script tries to retrieve hwsku information from config_db.
During initial boot up config_db might not be populated. In order to start topology service before config_db is updated,
update topology script to get hwsku information from minigraph.xml if it is available.
This will be helpful to bring up multi-asic VS testbed by loading minigraph and starting topology service.
- How I did it
Update topology.sh script to retrieve hwsku information from minigraph.xml.
Fix clean up function on msft_multi_asic_vs toplogy script.
- How to verify it
single-asic VS - no change; topology service is only enabled for multi-asic VS.
multi-asic VS - Bring up multi-asic VS image, copy minigraph to vs image, start topology service. Topology service should be successful.
to test clean up function fix, start topology service - make sure interfaces are created and moved to the right namespaces.
stop topology service - make sure namespace do not have any interface and all front end interfaces are present in default namespace.
What I did:-
For multi-asic platforms added iptable v4 rule to communicate on docker bridge ip
For multi-asic platforms extend iptable v4 rule for iptable v6 also
For multi-asic program made all internal rules applicable for all protocols (not filter based on tcp/udp). This is done to be consistent same as local host rule
For multi-asic platforms made nat rule (to forward traffic from namespace to host) generic for all protocols and also use Source IP if present for matching
This PR is port of #6845 for 201911
show ip interfaces is enhanced recently to support multi ASIC platforms in this Azure/sonic-utilities#1437. The ipintutil script as to run as sudo user, to get the ip interface from each namespace.
Add this script to the sudoer file so that show ip interface command is available for user with read-only permissions
Signed-off-by: Arvindsrinivasan Lakshmi Narasimhan <arlakshm@microsoft.com>
Signed-off-by: Arvindsrinivasan Lakshmi Narasimhan <arlakshm@microsoft.com>
Install pyroute2 for sonic-utilities. This change is needed for Azure/sonic-utilities#1437
#### Why I did it
Add new SKU for SN2700 Mellanox system that supports the following port configuration:
8 X 100G
40 X 50G
8 X 10G
#### How I did it
Add new Folder - "Mellanox-SN2700-D40C8S8" under /sonic-buildimage/device/mellanox/x86_64-mlnx_msn2700-r0/
that contains the relevant files supporting this SKU
#### How to verify it
Bring up the image, run "show interface status" and make sure that all ports are up and reflect the following requirement:
Port 1/3 will be used as 4x10G
Port 2/4 - Not exist (blocked since 1 and 3 split to 4)
Port 7/8/9/10/23/24/25/26 will used as 100G
All other ports will be used as 2x50G
#### Which release branch to backport (provide reason below if selected)
- [ ] 201811
- [X] 201911
- [ ] 202006
- [ ] 202012
#### Description for the changelog
Support new SKU under the name of SN2700-D40C8S8
In rare case can see that xcvrd failed due to "UnboundLocalError: local variable 'label_port' referenced before assignment"
Init "label_port" as None at the beginning of the function, to avoid the case that "label_port" not assigned.
It's been reported that accton fan monitor process keeps consuming memory after few days.
The amount of memory occupied increases in linear and never leased.
Signed-off-by: roy_lee <roy_lee@edge-core.com>
**- Why I did it**
- The thermalctld daemon on the Pmon docker requires support from the thermal manager API.
**- How I did it**
- Removed the old function for detecting a faulty fan.
- Removed the old function for detecting excess temperature.
- Implement thermal_manager APIs based on ThermalManagerBase
- Implement thermal_conditions APIs based on ThermalPolicyConditionBase
- Implement thermal_actions APIs based on ThermalPolicyActionBase
- Implement thermal_info APIs based on ThermalPolicyInfoBase
- Add thermal_policy.json
[multi_asic][vs]: Add dependency in teamd service to start after topology service.
- Why I did it
In multi-asic VS, topology service is run after database service to set up the internal asic topology.
swss and syncd have a dependency to start after topology service is run so that the interfaces are moved to right namespace and created in the right namespace. In case of multi-asic vs, during the initial boot up, when there is no configuration added, teamd service starts and swss/syncd do not start as topology service does not start. Upon loading configuration using config_db or minigraph, swss and sycnd start up , but teamd is not restarted as swss is not stopped and started. This causes teamd to be in a bad state and requires a reload of config.
- How I did it
Add dependency in teamd service to start after topology service is completed.
- How to verify it
No change in single asic vs or platform.
No change in multi-asic regular image.
Change only in multi-asic VS. Bring up a multi-asic VS image without any configration, teamd service will fail to start due to dependency failure. Load minigraph, start topology service, load configuration, ensure all services come up.
Signed-off-by: SuvarnaMeenakshi <sumeenak@microsoft.com>
- Introduced TS common file in docker as well and moved common functions.
- TSA/B/C scripts run only in BGP instances for front end ASICs.
In addition skip enforcing it on route maps used between internal BGP sessions.
admin@str--acs-1:~$ sudo /usr/bin/TSA
System Mode: Normal -> Maintenance
and in case of Multi-ASIC
admin@str--acs-1:~$ sudo /usr/bin/TSA
BGP0 : System Mode: Normal -> Maintenance
BGP1 : System Mode: Normal -> Maintenance
BGP2 : System Mode: Normal -> Maintenance
When large number of port channels (more than 64) is configured, a config reload command might ends with not all port channel configured and up. Further debug shows that unloading the port channels on the ASIC driver take a lot of time.
With the change, deleting all port channels before the syncd restart will free resources better and the ASIC driver will unload all netdev fast and the operation will execute properly.
Bring in a fix for thermal shutdown observed while executing warm-reboot:
- All | prevent FW access during ISSU
Signed-off-by: Stepan Blyschak <stepanb@nvidia.com>
Fix the build failure of asyncsnmp-2.1.0-py3-none-any.whl .
Looks like a recent update on https://packages.debian.org/stretch/libpython3.5-stdlib
3.5.3-1+deb9u1 -> 3.5.3-1+deb9u3 break the build command python3 setup.py bdist_wheel
Signed-off-by: Abhishek Dosi <abdosi@microsoft.com>
