* Build a Docker base image based on Debian Stretch
* Build a config-engine Docker image based on Stretch
* Do not install socat from Debian repo
* Add changes that were made to docker-base since this PR was opened
* QoS config change: 1) DSCP mapping; 2) link pg/queue 6 to lossy buffer;
3) redistribute scheduler
Signed-off-by: Wenda <wenni@microsoft.com>
* Add scheduling weight to queue 2
Signed-off-by: Wenda <wenni@microsoft.com>
* Link pg/queue 2 to lossy buffer
Signed-off-by: Wenda <wenni@microsoft.com>
* Update the pg headroom for a7060-D48C8 50G
Signed-off-by: Wenda <wenni@microsoft.com>
* Update config gen test for qos
Signed-off-by: Wenda <wenni@microsoft.com>
* Update pg headroom size, and update egress lossy pool size accordingly
Signed-off-by: Wenda <wenni@microsoft.com>
* Update headroom pool size; Update ingress service pool and egress lossy
pool sizes accordingly;
Signed-off-by: Wenda <wenni@microsoft.com>
* a7260: update headroom pool size; Update ingress service pool and egress lossy pool sizes accordingly;
Signed-off-by: Wenda <wenni@microsoft.com>
* Update config gen test for buffer
Signed-off-by: Wenda <wenni@microsoft.com>
* [reboot cause] Move reboot-cause files to /host directory so they persist across SONiC upgrades
* [sonic-utilities] Update submodule to include related changes
The sai profile itself can support 32x50G+16x100G/40G while
the initial port_config.ini uses 40G speeds for port 17-32.
Signed-off-by: Zhenggen Xu <zxu@linkedin.com>
- Also initialized the number of cos queues in AS5712 to 8.
Else the interfaces are not coming up due to a lower default value of cos queues in AS5712.
Fix:
- Enable alpm settings in the platform config file for AS5712.
- With the current values of l2_mem_entries (163840) and l3_mem_entries (81920), the memory is not sufficient. Scale them down to the
values that work for ALPM in as5712 [l2_mem_entries = 32768, l3_mem_entries = 16384].
Signed-off-by: kiran.kella@broadcom.com
- What I did
This fix removes the possibility of 'localhost' entry getting removed from /etc/hosts file by hostname-config service.
Without this change, whenever we change the hostname from 'localhost' to any other name on the config_db.json and reload the config, /etc/hosts file will only have the new hostname on it. But there are multiple sonic utilities (eg: swssconfig) which relies on the hard coded 'localhost' name and they tend to stop working.
- How I did it
Added a new check on hostname-config.sh script to avid blindly deleting the line containing the old hostname from /etc/hosts file. Now it will delete the old hostname only if its not localhost or when the hostname is not changing.
- How to verify it
Bring up SONiC on a device with hostname as localhost
Edit /etc/sonic/config_db.json to update the 'hostname' filed under DEVICE_METADATA from "hostname" : "localhost" --> "hostname" : "sonic"
run config reload -y to reflect the hostname change done on config_db.json file.
cat /etc/hosts and check whether both 127.0.0.1 localhost and 127.0.0.1 sonic entry are present on the file.
ping localhost should work fine.
- Description for the changelog
Make hostname-config service more robust in handling SONiC hostname change from localhost to anything else.
To include following changes:
- [teammgrd]: Add retry logic for starting port channel with teamd (#756)
- [portsorch] fix bug in initializePort (#753)
- [intfmgrd] Fix intfmgrd hanging untill first interface becomes ready (#748)
- [intfmgrd]: Support loopback (#742)
-Improve comments for neighbor warmrestart related functions and warmRestartAssist class (#740)
Signed-off-by: Ying Xie <ying.xie@microsoft.com>
When rebooting without the platform_reboot plugin, systemd takes a few
minutes to properly shutdown. It's blocking on some docker cleanup
operation.
As described by https://github.com/docker/for-linux/issues/421 there
is a race between docker.service and containerd.service.
docker needs containerd to properly stop the containers.
The race condition could happen like this:
When an interface is enslaved into the port channel immediately after
it is created, the order of creating the ifinfo and linking the ifinfo to
the port is not guaranteed.
Please check the patch commit message to get full details.
Signed-off-by: Shu0T1an ChenG <shuche@microsoft.com>
* Add psud daemon to pmon container
* Update submodule sonic-platform-daemons
Submodule update sonic-platform-daemons:
e5d8155 - [sonic-psud] add a new daemon sonic-psud to platform monitor (#20)
Signed-off-by: Kevin Wang <kevinw@mellanox.com>
dell_ich module fails to load sometimes due to the failure of pci_get_drvdata().
This function is responsible for fetching INTEL PCI related memory handle in kernel. This is implemented in lpc_ich kernel module.
Due to race in addition/deletion of kernel modules, sometimes lpc_ich loads after dell_ich.
Because of this behaviour dell_ich module fails to load.
Fixed by addding dependency between modules.
Removed i2c_mux_gpio module from blacklist entry as it is not the original root case of this issue.
This commit has the forward porting of changes from 20180330
to master. Unit tested by checking the broadcom registers
for the values. The values in hardware reflect the correct values
except for TC to PFC priority group map values. Master branch values
for TC to PFC priority group map for both z9100 and s6100 in incorrect,
I have a build of Oct 12 master and the values are correct in
that build.
Signed-off-by: Harish Venkatraman <Harish_Venkatraman@Dell.com>
