[GH-ISSUE #235] Mgmtd failed to listen local 8000 port #143

New issue

Closed

opened 2026-10-26 09:45:53 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-26 09:45:53 +00:00

Owner

Originally created by @SimonCqk on GitHub (Apr 9, 2025).
Original GitHub issue: https://github.com/deepseek-ai/3FS/issues/235

Hi all, I'm attempting to deploy 3fs on Kubernetes. While starting the mgmtd container, I noticed that the localhost is not listening on port 8000 (though IB devices are properly recognized). Here are the logs. There is a suspicious log entry mentioning missing zone information – should this be a concern?

025-04-09T03:54:54.125110083+00:00 mgmtd_main:   29 ServerLauncher.h:39 INFO] trace_format = 'json'
/root/.profile: line 10: /.cargo/env: No such file or directory
[2025-04-09T03:54:54.248634157+00:00 mgmtd_main:   29 IBDevice.cc:169 INFO] ibdev2netdev: mlx5_21 port 1 ==> eth0 (Up)
[2025-04-09T03:54:54.248651138+00:00 mgmtd_main:   29 IBDevice.cc:186 INFO] ibdev2netdev parsed: mlx5_21 => eth0
[2025-04-09T03:54:54.248860268+00:00 mgmtd_main:   29 IfAddrs.h:102 INFO] Get ifaddr of eth0, addr 33.200.166.55/24, subnet 33.200.166.0/24, up true
[2025-04-09T03:54:54.253876635+00:00 mgmtd_main:   29 IBDevice.cc:396 INFO] IBDevice mlx5_21:1 IP addrs 33.200.166.55/24
**[2025-04-09T03:54:54.253888711+00:00 mgmtd_main:   29 IBDevice.cc:441 CRITICAL] IBDevice mlx5_21:1 can't set zone by IP, fallback to UNKNOWN**
[2025-04-09T03:54:54.253951241+00:00 mgmtd_main:   29 IBDevice.cc:367 INFO] IBDevice mlx5_21 add active port 1, linklayer ETHERNET, addrs 33.200.166.55/24, zones UNKNOWN, RoCE v2 GID 0:0:0:0:0:0:0:0:0:0:ff:ff:21:c8:a6:37
[2025-04-09T03:54:54.253957300+00:00 mgmtd_main:   29 IBDevice.cc:256 INFO] IBDevice add mlx5_21, id 0, 1 available ports
[2025-04-09T03:54:54.255900011+00:00 mgmtd_main:   29 ServerLauncher.h:43 INFO] IBDevice inited
[2025-04-09T03:54:54.256000006+00:00 IBManager:  190 EventLoop.cc:116 INFO] EventLoop::loop() started.
[2025-04-09T03:54:54.263687018+00:00 mgmtd_main:   29 Utils.cc:239 INFO] Start load AppInfo from launcher

Originally created by @SimonCqk on GitHub (Apr 9, 2025). Original GitHub issue: https://github.com/deepseek-ai/3FS/issues/235 Hi all, I'm attempting to deploy 3fs on Kubernetes. While starting the mgmtd container, I noticed that the localhost **is not listening on port 8000** (though IB devices are properly recognized). Here are the logs. There is a suspicious log entry mentioning missing zone information – should this be a concern? ``` 025-04-09T03:54:54.125110083+00:00 mgmtd_main: 29 ServerLauncher.h:39 INFO] trace_format = 'json' /root/.profile: line 10: /.cargo/env: No such file or directory [2025-04-09T03:54:54.248634157+00:00 mgmtd_main: 29 IBDevice.cc:169 INFO] ibdev2netdev: mlx5_21 port 1 ==> eth0 (Up) [2025-04-09T03:54:54.248651138+00:00 mgmtd_main: 29 IBDevice.cc:186 INFO] ibdev2netdev parsed: mlx5_21 => eth0 [2025-04-09T03:54:54.248860268+00:00 mgmtd_main: 29 IfAddrs.h:102 INFO] Get ifaddr of eth0, addr 33.200.166.55/24, subnet 33.200.166.0/24, up true [2025-04-09T03:54:54.253876635+00:00 mgmtd_main: 29 IBDevice.cc:396 INFO] IBDevice mlx5_21:1 IP addrs 33.200.166.55/24 **[2025-04-09T03:54:54.253888711+00:00 mgmtd_main: 29 IBDevice.cc:441 CRITICAL] IBDevice mlx5_21:1 can't set zone by IP, fallback to UNKNOWN** [2025-04-09T03:54:54.253951241+00:00 mgmtd_main: 29 IBDevice.cc:367 INFO] IBDevice mlx5_21 add active port 1, linklayer ETHERNET, addrs 33.200.166.55/24, zones UNKNOWN, RoCE v2 GID 0:0:0:0:0:0:0:0:0:0:ff:ff:21:c8:a6:37 [2025-04-09T03:54:54.253957300+00:00 mgmtd_main: 29 IBDevice.cc:256 INFO] IBDevice add mlx5_21, id 0, 1 available ports [2025-04-09T03:54:54.255900011+00:00 mgmtd_main: 29 ServerLauncher.h:43 INFO] IBDevice inited [2025-04-09T03:54:54.256000006+00:00 IBManager: 190 EventLoop.cc:116 INFO] EventLoop::loop() started. [2025-04-09T03:54:54.263687018+00:00 mgmtd_main: 29 Utils.cc:239 INFO] Start load AppInfo from launcher ```

bsakaguchi closed this issue

2026-10-26 09:45:54 +00:00

bsakaguchi commented

2026-10-26 09:45:54 +00:00

Author

Owner

@bugwz commented on GitHub (Apr 11, 2025):

The message can't set zone by IP, fallback to UNKNOWN** also appears in my deployment cluster but doesn't affect its operation.

I believe it's simply checking whether a set of addresses belong to the subnets defined in the configuration to determine network zones. If no addresses match the subnets, it will attempt to use either the default network zone from the configuration or the zone specified by environment variables. Therefore, this prompt is not an error.

@bugwz commented on GitHub (Apr 11, 2025): The message `can't set zone by IP, fallback to UNKNOWN**` also appears in my deployment cluster but doesn't affect its operation. I believe it's simply checking whether a set of addresses belong to the subnets defined in the configuration to determine network zones. If no addresses match the subnets, it will attempt to use either the default network zone from the configuration or the zone specified by environment variables. Therefore, this prompt is not an error.

bsakaguchi commented

2026-10-26 09:45:55 +00:00

Author

Owner

@SimonCqk commented on GitHub (Apr 13, 2025):

The root cause is foundationdb not ready, and mgmtd itself hang because of that, the issue has fixed and close it.

@SimonCqk commented on GitHub (Apr 13, 2025): The root cause is foundationdb not ready, and mgmtd itself hang because of that, the issue has fixed and close it.

bsakaguchi referenced this issue

2026-10-26 09:48:32 +00:00

[PR #143] Resolve mgmtd address name #308