Master backup#1
Draft
matiAlfaro wants to merge 5 commits into
Draft
Conversation
added 5 commits
February 22, 2018 11:49
amazor
pushed a commit
that referenced
this pull request
Feb 19, 2024
…bors over iBGP Session (sonic-net#16705) What I did: Enable Sending BGP Community over internal neighbors over iBGP Session Microsoft ADO: 25268695 Why I did: Without this change BGP community send by e-BGP Peers are not carry-forward to other e-BGP peers. str2-xxxx-lc1-2# show bgp ipv6 20c0:a801::/64 BGP routing table entry for 20c0:a801::/64, version 52141 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65500 2603:10e2:400::6 from 2603:10e2:400::6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) Last update: Tue Sep 26 16:08:26 2023 str2-xxxx-lc1-2# show ip bgp 192.168.35.128/25 BGP routing table entry for 192.168.35.128/25, version 52688 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65502 3.3.3.6 from 3.3.3.6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) Last update: Tue Sep 26 15:45:51 2023 After the change str2-xxxx-lc2-2(config)# router bgp 65100 str2-xxxx-lc2-2(config-router)# address-family ipv4 str2-xxxx-lc2-2(config-router-af)# neighbor INTERNAL_PEER_V4 send-community str2-xxxx-lc2-2(config-router-af)# exit str2-xxxx-lc2-2(config-router)# address-family ipv6 str2-xxxx-lc2-2(config-router-af)# neighbor INTERNAL_PEER_V6 send-community str2-xxxx-lc1-2# show bgp ipv6 20c0:a801::/64 BGP routing table entry for 20c0:a801::/64, version 52400 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65500 2603:10e2:400::6 from 2603:10e2:400::6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) **Community: 1111:1111** Last update: Tue Sep 26 16:10:19 2023 str2-xxxx-lc1-2# show ip bgp 192.168.35.128/25 BGP routing table entry for 192.168.35.128/25, version 52947 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65502 3.3.3.6 from 3.3.3.6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) **Community: 1111:1111** Last update: Tue Sep 26 16:10:09 2023 Signed-off-by: Abhishek Dosi <abdosi@microsoft.com>
amazor
pushed a commit
that referenced
this pull request
Apr 30, 2024
…bors over iBGP Session (sonic-net#16705) What I did: Enable Sending BGP Community over internal neighbors over iBGP Session Microsoft ADO: 25268695 Why I did: Without this change BGP community send by e-BGP Peers are not carry-forward to other e-BGP peers. str2-xxxx-lc1-2# show bgp ipv6 20c0:a801::/64 BGP routing table entry for 20c0:a801::/64, version 52141 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65500 2603:10e2:400::6 from 2603:10e2:400::6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) Last update: Tue Sep 26 16:08:26 2023 str2-xxxx-lc1-2# show ip bgp 192.168.35.128/25 BGP routing table entry for 192.168.35.128/25, version 52688 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65502 3.3.3.6 from 3.3.3.6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) Last update: Tue Sep 26 15:45:51 2023 After the change str2-xxxx-lc2-2(config)# router bgp 65100 str2-xxxx-lc2-2(config-router)# address-family ipv4 str2-xxxx-lc2-2(config-router-af)# neighbor INTERNAL_PEER_V4 send-community str2-xxxx-lc2-2(config-router-af)# exit str2-xxxx-lc2-2(config-router)# address-family ipv6 str2-xxxx-lc2-2(config-router-af)# neighbor INTERNAL_PEER_V6 send-community str2-xxxx-lc1-2# show bgp ipv6 20c0:a801::/64 BGP routing table entry for 20c0:a801::/64, version 52400 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65500 2603:10e2:400::6 from 2603:10e2:400::6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) **Community: 1111:1111** Last update: Tue Sep 26 16:10:19 2023 str2-xxxx-lc1-2# show ip bgp 192.168.35.128/25 BGP routing table entry for 192.168.35.128/25, version 52947 Paths: (1 available, best #1, table default) Not advertised to any peer 65000 65502 3.3.3.6 from 3.3.3.6 (3.3.3.6) Origin IGP, localpref 100, valid, internal, best (First path received) **Community: 1111:1111** Last update: Tue Sep 26 16:10:09 2023 Signed-off-by: Abhishek Dosi <abdosi@microsoft.com>
amazor
pushed a commit
that referenced
this pull request
Oct 1, 2024
#### Why I did it
Dropping control character (message sent when XSUB connects to XPUB as part of ZMQ Proxy setup to notify that subscription has been made) in do capture has been flaky since control character is not guaranteed to be the first message sent if there are events (like event-down-ctr) being published to XSUB.
Scenarios
1) Control character is sent and is first message when starting capture service
`eventd#eventd#eventd: :- heartbeat_ctrl: Set heartbeat_ctrl pause=1`
`eventd#eventd#eventd: :- do_capture: Received subscription message when XSUB connects to XPUB`
2) Events like event-down ctr is sent before control character
`eventd#eventd#eventd: :- run: Dropping Message: 22 serialization::archive 18 17 sonic-events-host`
`eventd#eventd#eventd: :- run: Dropping Message: 22 serialization::archive 18 0 0 4 0 0 0 1 d 103 {"sonic-events-host:event-stopped-ctr":{"ctr_name":"EVENTD","timestamp":"2024-08-27T00:02:51.407518Z"}} 1 r 36 3357542f-bae1-458f-a804-660e620d21f5 1 s 1 9 1 t 19 1724716971407591080`
`heartbeat_ctrl: Set heartbeat_ctrl pause=1`
`do_capture: Received subscription message when XSUB connects to XPUB`
3) Control character is not sent at all
`eventd#eventd#eventd: :- heartbeat_ctrl: Set heartbeat_ctrl pause=1`
4) Control character is delayed and not caught when starting capture service, but is then caught after causing deserialize error.
`do_capture: Receiving event from source: 22 serialization::archive 18 17 sonic-events-host, will read second part of event`
`deserialize: deserialize Failed: input stream errorstr[0:64]:(#1) data type: std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&`
`zmq_read_part: Failed to deserialize part rc=-2`
`zmq_read_part: last:errno=11`
`zmq_message_read: Failure to read part1 rc=-2`
`zmq_message_read: last:errno=11`
We can cover these scenarios by just dropping the control character inside zmq_message_read as part of events_common in swsscommon (different PR). In this PR we will remove such handling logic and make sure that empty events that will be sent by control character are ignored.
##### Work item tracking
- Microsoft ADO **(number only)**:28728116
#### How I did it
Remove logic for handling control character
#### How to verify it
UT and sonic-mgmt test cases.
VladimirKuk
pushed a commit
that referenced
this pull request
Nov 5, 2024
#### Why I did it
Dropping control character (message sent when XSUB connects to XPUB as part of ZMQ Proxy setup to notify that subscription has been made) in do capture has been flaky since control character is not guaranteed to be the first message sent if there are events (like event-down-ctr) being published to XSUB.
Scenarios
1) Control character is sent and is first message when starting capture service
`eventd#eventd#eventd: :- heartbeat_ctrl: Set heartbeat_ctrl pause=1`
`eventd#eventd#eventd: :- do_capture: Received subscription message when XSUB connects to XPUB`
2) Events like event-down ctr is sent before control character
`eventd#eventd#eventd: :- run: Dropping Message: 22 serialization::archive 18 17 sonic-events-host`
`eventd#eventd#eventd: :- run: Dropping Message: 22 serialization::archive 18 0 0 4 0 0 0 1 d 103 {"sonic-events-host:event-stopped-ctr":{"ctr_name":"EVENTD","timestamp":"2024-08-27T00:02:51.407518Z"}} 1 r 36 3357542f-bae1-458f-a804-660e620d21f5 1 s 1 9 1 t 19 1724716971407591080`
`heartbeat_ctrl: Set heartbeat_ctrl pause=1`
`do_capture: Received subscription message when XSUB connects to XPUB`
3) Control character is not sent at all
`eventd#eventd#eventd: :- heartbeat_ctrl: Set heartbeat_ctrl pause=1`
4) Control character is delayed and not caught when starting capture service, but is then caught after causing deserialize error.
`do_capture: Receiving event from source: 22 serialization::archive 18 17 sonic-events-host, will read second part of event`
`deserialize: deserialize Failed: input stream errorstr[0:64]:(#1) data type: std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&`
`zmq_read_part: Failed to deserialize part rc=-2`
`zmq_read_part: last:errno=11`
`zmq_message_read: Failure to read part1 rc=-2`
`zmq_message_read: last:errno=11`
We can cover these scenarios by just dropping the control character inside zmq_message_read as part of events_common in swsscommon (different PR). In this PR we will remove such handling logic and make sure that empty events that will be sent by control character are ignored.
##### Work item tracking
- Microsoft ADO **(number only)**:28728116
#### How I did it
Remove logic for handling control character
#### How to verify it
UT and sonic-mgmt test cases.
shiraez
pushed a commit
that referenced
this pull request
Dec 11, 2024
…et#21095) Adding the below fix from FRR FRRouting/frr#17297 This is to fix the following crash which is a statistical issue [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Core was generated by `/usr/lib/frr/zebra -A 127.0.0.1 -s 90000000 -M dplane_fpm_nl -M snmp'. Program terminated with signal SIGABRT, Aborted. #0 0x00007fccd7351e2c in ?? () from /lib/x86_64-linux-gnu/libc.so.6 [Current thread is 1 (Thread 0x7fccd6faf7c0 (LWP 36))] (gdb) bt #0 0x00007fccd7351e2c in ?? () from /lib/x86_64-linux-gnu/libc.so.6 #1 0x00007fccd7302fb2 in raise () from /lib/x86_64-linux-gnu/libc.so.6 #2 0x00007fccd72ed472 in abort () from /lib/x86_64-linux-gnu/libc.so.6 #3 0x00007fccd75bb3a9 in _zlog_assert_failed (xref=xref@entry=0x7fccd7652380 <_xref.16>, extra=extra@entry=0x0) at ../lib/zlog.c:678 #4 0x00007fccd759b2fe in route_node_delete (node=<optimized out>) at ../lib/table.c:352 #5 0x00007fccd759b445 in route_unlock_node (node=0x0) at ../lib/table.h:258 #6 route_next (node=<optimized out>) at ../lib/table.c:436 #7 route_next (node=node@entry=0x56029d89e560) at ../lib/table.c:410 #8 0x000056029b6b6b7a in if_lookup_by_name_per_ns (ns=ns@entry=0x56029d873d90, ifname=ifname@entry=0x7fccc0029340 "PortChannel1020") at ../zebra/interface.c:312 #9 0x000056029b6b8b36 in zebra_if_dplane_ifp_handling (ctx=0x7fccc0029310) at ../zebra/interface.c:1867 #10 zebra_if_dplane_result (ctx=0x7fccc0029310) at ../zebra/interface.c:2221 #11 0x000056029b7137a9 in rib_process_dplane_results (thread=<optimized out>) at ../zebra/zebra_rib.c:4810 #12 0x00007fccd75a0e0d in thread_call (thread=thread@entry=0x7ffe8e553cc0) at ../lib/thread.c:1990 #13 0x00007fccd7559368 in frr_run (master=0x56029d65a040) at ../lib/libfrr.c:1198 sonic-net#14 0x000056029b6ac317 in main (argc=9, argv=0x7ffe8e5540d8) at ../zebra/main.c:478
VladimirKuk
pushed a commit
that referenced
this pull request
Jan 21, 2025
…et#21095) Adding the below fix from FRR FRRouting/frr#17297 This is to fix the following crash which is a statistical issue [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Core was generated by `/usr/lib/frr/zebra -A 127.0.0.1 -s 90000000 -M dplane_fpm_nl -M snmp'. Program terminated with signal SIGABRT, Aborted. #0 0x00007fccd7351e2c in ?? () from /lib/x86_64-linux-gnu/libc.so.6 [Current thread is 1 (Thread 0x7fccd6faf7c0 (LWP 36))] (gdb) bt #0 0x00007fccd7351e2c in ?? () from /lib/x86_64-linux-gnu/libc.so.6 #1 0x00007fccd7302fb2 in raise () from /lib/x86_64-linux-gnu/libc.so.6 #2 0x00007fccd72ed472 in abort () from /lib/x86_64-linux-gnu/libc.so.6 #3 0x00007fccd75bb3a9 in _zlog_assert_failed (xref=xref@entry=0x7fccd7652380 <_xref.16>, extra=extra@entry=0x0) at ../lib/zlog.c:678 #4 0x00007fccd759b2fe in route_node_delete (node=<optimized out>) at ../lib/table.c:352 #5 0x00007fccd759b445 in route_unlock_node (node=0x0) at ../lib/table.h:258 #6 route_next (node=<optimized out>) at ../lib/table.c:436 #7 route_next (node=node@entry=0x56029d89e560) at ../lib/table.c:410 #8 0x000056029b6b6b7a in if_lookup_by_name_per_ns (ns=ns@entry=0x56029d873d90, ifname=ifname@entry=0x7fccc0029340 "PortChannel1020") at ../zebra/interface.c:312 #9 0x000056029b6b8b36 in zebra_if_dplane_ifp_handling (ctx=0x7fccc0029310) at ../zebra/interface.c:1867 #10 zebra_if_dplane_result (ctx=0x7fccc0029310) at ../zebra/interface.c:2221 #11 0x000056029b7137a9 in rib_process_dplane_results (thread=<optimized out>) at ../zebra/zebra_rib.c:4810 #12 0x00007fccd75a0e0d in thread_call (thread=thread@entry=0x7ffe8e553cc0) at ../lib/thread.c:1990 #13 0x00007fccd7559368 in frr_run (master=0x56029d65a040) at ../lib/libfrr.c:1198 sonic-net#14 0x000056029b6ac317 in main (argc=9, argv=0x7ffe8e5540d8) at ../zebra/main.c:478
apannerselva
pushed a commit
that referenced
this pull request
Feb 19, 2026
#### Why I did it If one python wheel is already installed inside slave container, it will not install again. Below is a sample log: ``` sed: -e expression #1, char 11: extra characters after command WARNING: The directory '/var/user/.cache/pip' or its parent directory is not owned or is not writable by the current user. The cache has been disabled. Check the permissions and owner of that directory. If executing pip with sudo, you should use sudo's -H flag. Processing ./target/python-wheels/bookworm/sonic_yang_models-1.0-py3-none-any.whl sonic-yang-models is already installed with the same version as the provided wheel. Use --force-reinstall to force an installation of the wheel. WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager, possibly rendering your system unusable.It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what you are doing and want to suppress this warning. [notice] A new release of pip is available: 24.2 -> 25.3 [notice] To update, run: python3 -m pip install --upgrade pip Build end time: Wed Dec 3 22:53:07 UTC 2025 Elapsed time: 0h 0m 1s ``` However, we expect to reinstall the python wheel for target `$(PYTHON_WHEELS_PATH)/%-install` ##### Work item tracking - Microsoft ADO **(number only)**: #### How I did it Update slave.mk to enasure force install the python wheel. #### How to verify it After this change, local build will successfully force install the python wheel. See new logs: ``` sed: -e expression #1, char 11: extra characters after command WARNING: The directory '/var/qiluo/.cache/pip' or its parent directory is not owned or is not writable by the current user. The cache has been disabled. Check the permissions and owner of that directory. If executing pip with sudo, you should use sudo's -H flag. Processing ./target/python-wheels/bookworm/sonic_yang_models-1.0-py3-none-any.whl Installing collected packages: sonic-yang-models Attempting uninstall: sonic-yang-models Found existing installation: sonic-yang-models 1.0 Uninstalling sonic-yang-models-1.0: Successfully uninstalled sonic-yang-models-1.0 Successfully installed sonic-yang-models-1.0 WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager, possibly rendering your system unusable.It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what you are doing and want to suppress this warning. [notice] A new release of pip is available: 24.2 -> 25.3 [notice] To update, run: python3 -m pip install --upgrade pip Build end time: Wed Dec 3 23:59:31 UTC 2025 ```
apannerselva
pushed a commit
that referenced
this pull request
Feb 19, 2026
…logs The `imklog` plugin of rsyslog collects the kernel logs from `/dev/kmsg` and enqueues it to the syslog. With `CONFIG_PRINTK_TIME` the kernel messages are by default prefixed with the elapsed time since boot. The `imklog` plugin parsing these messages have a few options such as to keep the timestamps as such or to interpret and adjust the syslog's reported time accordingly. The rsylog release `8.2312.0` has fixes in interpreting these timestamps, leading to the change in behavior observed in sonic-net#24386. https://salsa.debian.org/debian/rsyslog/-/blob/debian/8.2504.0-1/ChangeLog?ref_type=tags#L619 To restore the earlier behavior or retaining the kernel reported elapsed time, disable `KlogParseKernelTimestamp` as this leads to removal of timestamp from kernel messages and enable `KlogKeepKernelTimestamp` explicitly. The later is required as the default is now to discard the kernel timestamp. With this change, the logs retain the kernel timestamp: root@sonic:~# cat /var/log/syslog | grep "sonic.*kernel:" | head -n 3 2025 Nov 4 05:15:14.918946 sonic NOTICE kernel: [ 0.000000] Linux version 6.12.41+deb13-sonic-amd64 (debian-kernel@lists.debian.org) (x86_64-linux-gnu-gcc-14 (Debian 14.2.0-19) 14.2.0, GNU ld (GNU Binutils for Debian) 2.44) #1 SMP PREEMPT_DYNAMIC Debian 6.12.41-1 (2025-08-12) 2025 Nov 4 05:15:14.919533 sonic INFO kernel: [ 0.000000] Command line: BOOT_IMAGE=/image-trixie.0-dirty-20251102.122837/boot/vmlinuz-6.12.41+deb13-sonic-amd64 root=UUID=ac0b6826-f8a3-461f-a8ff-701df60d90b6 rw console=tty0 console=ttyS0,115200n8 quiet processor.max_cstate=1 intel_idle.max_cstate=0 net.ifnames=0 biosdevname=0 loop=image-trixie.0-dirty-20251102.122837/fs.squashfs loopfstype=squashfs apparmor=1 security=apparmor varlog_size=4096 usbcore.autosuspend=-1 intel_iommu=off modprobe.blacklist=gpio_ich,i2c-ismt,i2c_ismt,i2c-i801,i2c_i801 crashkernel=0M-2G:256M,2G-4G:320M,4G-8G:384M,8G-:448M acpi_no_watchdog 2025 Nov 4 05:15:14.919536 sonic INFO kernel: [ 0.000000] BIOS-provided physical RAM map: root@sonic:~# cat /var/log/syslog | grep "sonic.*kernel:" | tail -n 3 2025 Nov 4 05:17:26.831607 sonic WARNING kernel: [ 143.527486] PDDF_LED set_status_led: Set [FANTRAY_LED;1] color[green] 2025 Nov 4 05:17:26.912442 sonic WARNING kernel: [ 143.607086] PDDF_LED set_status_led: Set [FANTRAY_LED;2] color[green] 2025 Nov 4 05:20:32.499634 sonic WARNING kernel: [ 329.195319] PDDF_LED set_status_led: Set [SYS_LED;0] color[amber] root@sonic:~# Signed-off-by: Ramasamy Chandramouli <rachandr@celestica.com> Co-authored-by: Ramasamy Chandramouli <rachandr@celestica.com>
apannerselva
pushed a commit
that referenced
this pull request
May 6, 2026
…net#25643) * [build] Add build timing report and dependency analysis tools Add three scripts for build performance instrumentation: - scripts/build-timing-report.sh: Parse per-package timing from build logs (HEADER/FOOTER timestamps), generate sorted duration table, phase breakdown, parallelism timeline, and CSV export. - scripts/build-dep-graph.py: Parse rules/*.mk dependency graph, compute critical path, fan-out/fan-in bottleneck analysis, and generate DOT/JSON output for visualization. - scripts/build-resource-monitor.sh: Sample CPU, memory, disk I/O, and Docker container count during builds for resource utilization analysis. Add "make build-report" target to slave.mk that runs the timing report and dependency analysis after a build completes. Example output from a VS build on 24-core/30GB machine: - 210 packages built in 53m wall time (173m CPU) - Max concurrency: 5 (with SONIC_CONFIG_BUILD_JOBS=4) - Critical path: 14 packages deep (libnl -> libswsscommon -> utilities) - Top bottleneck: LIBSWSSCOMMON with 48 downstream dependents Signed-off-by: Rustiqly <rustiqly@users.noreply.github.com> * Address Copilot review: fix 17 bugs in build analysis scripts - Use free -m with division instead of free -g to avoid rounding (#1) - Add = and ?= to Makefile dependency regex patterns (#2, #7) - CPU calculation now uses /proc/stat delta (two reads) (#3, sonic-net#14) - Fix misleading 'critical path estimate' comment (#4) - Fix parallelism timeline comment (60s not 10s) (#5) - Include after-relationship packages in fan stats (#6) - Guard disk I/O division by zero when INTERVAL<=1 (#8) - Remove unused elapsed_line variable (#9) - Remove redundant LIBSWSSCOMMON_DBG check (#10) - Remove active_make_jobs from CSV header comment (#11) - Wire up _RDEPENDS parsing to build reverse deps (#12) - Remove unnecessary 'if v' filter on rdeps JSON (#13) - Remove unused REPORT_FORMAT parameter (sonic-net#15) - Add cycle detection to critical path algorithm (sonic-net#16) - Add execute permission check for companion scripts (sonic-net#17) Signed-off-by: Rustiqly <rustiqly@users.noreply.github.com> --------- Signed-off-by: Rustiqly <rustiqly@users.noreply.github.com> Co-authored-by: Rustiqly <rustiqly@users.noreply.github.com>
apannerselva
pushed a commit
that referenced
this pull request
May 6, 2026
…dating udevd rules (sonic-net#26343) - Why I did it On SONiC SmartSwitch platforms with DPUs, systemd-udevd crashes with SIGABRT on every reboot when DPU firmware initialization is slow. During the initramfs boot phase, a standalone systemd-udevd daemon is started to handle device discovery. If DPU firmware takes longer than the 60-second udevadm settle timeout (BlueField-3 DPUs can take 120 seconds each in the failure case when they are stuck), the initramfs cannot stop this udevd before switch_root. The stale process survives into the real system but is never chrooted into the overlayfs root, leaving it with a broken filesystem view. When dpu-udev-manager.sh writes udev rules, the stale udevd detects the change and crashes on an assertion in systemd's chase() path resolution (assert(path_is_absolute(p)) at chase.c:648), because dir_fd_is_root() returns false for a process whose root still points to the initramfs rootfs rather than the overlayfs. This triggers a systemd issue : systemd/systemd#29559 which maintainers doesn't consider as a bug from systemd side. Raising this fix for our usecase. Core was generated by `/usr/lib/systemd/systemd-udevd --daemon --resolve-names=never'. Program terminated with signal SIGABRT, Aborted. #0 0x00007f29fe7f695c in ?? () from /lib/x86_64-linux-gnu/libc.so.6 (gdb) bt #0 0x00007f29fe7f695c in ?? () from /lib/x86_64-linux-gnu/libc.so.6 #1 0x00007f29fe7a1cc2 in raise () from /lib/x86_64-linux-gnu/libc.so.6 #2 0x00007f29fe78a4ac in abort () from /lib/x86_64-linux-gnu/libc.so.6 #3 0x00007f29fea50c11 in ?? () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #4 0x00007f29feb94a8b in chase () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #5 0x00007f29feb956e2 in chase_and_opendir () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #6 0x00007f29feb9a609 in conf_files_list_strv () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #7 0x00007f29fea913e8 in config_get_stats_by_path () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #8 0x0000559f295519cf in ?? () #9 0x0000559f29553a77 in ?? () #10 0x00007f29fec36055 in ?? () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #11 0x00007f29fec3668d in sd_event_dispatch () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #12 0x00007f29fec394a8 in sd_event_run () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so #13 0x00007f29fec396c7 in sd_event_loop () from /usr/lib/x86_64-linux-gnu/systemd/libsystemd-shared-257.so sonic-net#14 0x0000559f29545820 in ?? () sonic-net#15 0x00007f29fe78bca8 in ?? () from /lib/x86_64-linux-gnu/libc.so.6 sonic-net#16 0x00007f29fe78bd65 in __libc_start_main () from /lib/x86_64-linux-gnu/libc.so.6 sonic-net#17 0x0000559f29545c51 in ?? () - How I did it Added a kill_stale_udevd() function to dpu-udev-manager.sh that runs before writing the udev rules. It identifies the systemd-managed udevd PID via systemctl show, then kills any other systemd-udevd --daemon process that doesn't match -- these are leftover initramfs instances. If no stale process exists (e.g. DPUs are healthy and the initramfs udevd exited cleanly), the function is a no-op. - How to verify it Deploy the image on a SmartSwitch with DPUs in a state where firmware initialization times out (>60s per DPU) by stopping image installation before firmware install step Reboot the switch Verify no new systemd-udevd coredumps in /var/core/ Verify the stale process was killed: journalctl -b 0 | grep dpu-udev-manager should show killing stale initramfs udevd PID (systemd udevd is PID ) Verify systemd-udevd.service is healthy: systemctl status systemd-udevd should show active (running) Verify DPU udev rules were written: cat /etc/udev/rules.d/92-midplane-intf.rules should contain the DPU interface naming rules Signed-off-by: Hemanth Kumar Tirupati <tirupatihemanthkumar@gmail.com>
apannerselva
pushed a commit
that referenced
this pull request
Jun 25, 2026
…26549) - Why I Did It Bug: Powering DPUs off at runtime via dpuctl leaves stale MST PCI device entries on the host. Subsequent mst status / mlxfwmanager calls then hang for ~50 seconds while MST tries to talk to the now-absent DPU PCI devices. Root cause: MST was started globally at boot via the mlnx-fw-manager.service systemd unit (ExecStartPre=/usr/bin/mst start --with_i2cdev) and left running. So MST always saw the DPU PCI devices, even after they were powered off, and any tool that walked them stalled. This was already fixed on 202511 — see sonic-net#25575 and sonic-net#26131. This PR ports the same fix to master/202605. - How I Did It Move MST lifecycle out of the global systemd unit and into the firmware-update paths that actually need MST. The host then runs without MST loaded, so DPU power-off no longer leaves stale entries that hang mst/mlxfwmanager. mlnx-fw-manager.service — Drop ExecStartPre=/usr/bin/mst start --with_i2cdev and ExecStop=/usr/bin/mst stop, so the unit no longer leaves MST running globally. Add an ExecCondition that skips the unit during SONIC_BOOT_TYPE=fastfast. mellanox_fw_manager (ASIC firmware upgrade) — FirmwareCoordinator now owns the MST lifecycle: _start_mst() runs mst start --with_i2cdev before the upgrade. _stop_mst() runs mst stop in a finally block, so MST is always stopped on exit. New ignore_mst_start_failure flag (CLI: -m / --ignore-mst-start-failure in main.py) lets callers continue when MST can't start — used by the BlueField installer where DPUs may be off. BlueField installer (install.sh.j2) — Remove the explicit chroot ... mst start and call mlnx-fw-manager -m --nosyslog --verbose so install doesn't abort if MST start fails. DPU FPGA upgrade (sonic_platform/component.py) — ComponentFPGADPU._install_firmware() now runs cpldupdate inside a new _mst_context() context manager that does mst start before and mst stop after, even on failure. ComponentCPLD.__get_mst_device() is also rewritten to use asic_detect.sh -p instead of scanning the MST device path, so it no longer depends on MST being globally loaded. Unit tests — Updated for the new __get_mst_device contract, the new ignore_mst_start_failure plumbing, and _start_mst/_stop_mst patching in coordinator tests. - How to Verify It 1. MST no longer loaded globally; DPU power-off doesn't hang host tools ✅ # dpuctl dpu-power-off --all ... # time mst status MST modules: ------------ MST PCI module is not loaded MST PCI configuration module is not loaded PCI Devices: ------------ 06:00.0 real 0m0.091s user 0m0.041s sys 0m0.068s # time mlxfwmanager Querying Mellanox devices firmware ... Device #1: ---------- Device Type: Spectrum3 ... real 0m0.176s user 0m0.013s sys 0m0.134s 2. ASIC firmware upgrade at boot still works ✅ mlnx-fw-manager starts MST, performs the upgrade, and stops MST. Verified end-to-end at boot time. 3. DPU FPGA upgrade still works ✅ cpldupdate runs inside the new _mst_context; MST is started/stopped only for the duration of the update. 4. BlueField installer ✅ Installer-time firmware upgrade succeeds with mlnx-fw-manager -m, including when DPUs are powered off and MST can't start. 5. Unit tests ✅ pytest platform/mellanox/mlnx-platform-api/tests/test_component.py pytest platform/mellanox/platform-utils/tests/ Signed-off-by: Yizhen Zhang <evazha@nvidia.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why I did it
Work item tracking
How I did it
How to verify it
Which release branch to backport (provide reason below if selected)
Tested branch (Please provide the tested image version)
Description for the changelog
Link to config_db schema for YANG module changes
A picture of a cute animal (not mandatory but encouraged)