Repo Timeout in China
Getting a lot of timeouts from nvidia repos. For example: https://developer.download.nvidia.com/compute/machine-learning/repos/rhel7/x86_64
Here's the log from my docker build, and the same process was fine yesterday:
#5 63.43 + sudo yum repolist
3449 #5 63.55 Loaded plugins: fastestmirror, ovl, priorities
3450 #5 63.64 Loading mirror speeds from cached hostfile
3451 #5 64.59 Retrieving key from https://nvidia.github.io/libnvidia-container/gpgkey
3452 #5 64.71 Importing GPG key 0xF796ECB0:
3453 #5 64.71 Userid : "NVIDIA CORPORATION (Open Source Projects) <cudatools@nvidia.com>"
3454 #5 64.71 Fingerprint: c95b 321b 61e8 8c18 09c4 f759 ddca e044 f796 ecb0
3455 #5 64.71 From : https://nvidia.github.io/libnvidia-container/gpgkey
3456 #5 64.71 https://nvidia.github.io/libnvidia-container/centos7/x86_64/repodata/repomd.xml: [Errno -1] Gpg Keys not imported, cannot verify repomd.xml for repo libnvidia-container
3457 #5 64.71 Trying other mirror.
3458 #5 68.07 Is this ok [y/N]: Retrieving key from https://nvidia.github.io/libnvidia-container/gpgkey
3459 #5 68.19 Importing GPG key 0xF796ECB0:
3460 #5 68.19 Userid : "NVIDIA CORPORATION (Open Source Projects) <cudatools@nvidia.com>"
3461 #5 68.19 Fingerprint: c95b 321b 61e8 8c18 09c4 f759 ddca e044 f796 ecb0
3462 #5 68.19 From : https://nvidia.github.io/libnvidia-container/gpgkey
3463 #5 68.19 https://nvidia.github.io/libnvidia-container/centos7/x86_64/repodata/repomd.xml: [Errno -1] Gpg Keys not imported, cannot verify repomd.xml for repo libnvidia-container
3464 #5 68.19 Trying other mirror.
3465 #5 68.42 Is this ok [y/N]: Retrieving key from https://nvidia.github.io/nvidia-container-runtime/gpgkey
3466 #5 68.55 Importing GPG key 0xF796ECB0:
3467 #5 68.55 Userid : "NVIDIA CORPORATION (Open Source Projects) <cudatools@nvidia.com>"
3468 #5 68.55 Fingerprint: c95b 321b 61e8 8c18 09c4 f759 ddca e044 f796 ecb0
3469 #5 68.55 From : https://nvidia.github.io/nvidia-container-runtime/gpgkey
3470 #5 68.55 https://nvidia.github.io/nvidia-container-runtime/centos7/x86_64/repodata/repomd.xml: [Errno -1] Gpg Keys not imported, cannot verify repomd.xml for repo nvidia-container-runtime
3471 #5 68.55 Trying other mirror.
3472 #5 68.79 Is this ok [y/N]: Retrieving key from https://nvidia.github.io/nvidia-docker/gpgkey
3473 #5 68.91 Importing GPG key 0xF796ECB0:
3474 #5 68.91 Userid : "NVIDIA CORPORATION (Open Source Projects) <cudatools@nvidia.com>"
3475 #5 68.91 Fingerprint: c95b 321b 61e8 8c18 09c4 f759 ddca e044 f796 ecb0
3476 #5 68.91 From : https://nvidia.github.io/nvidia-docker/gpgkey
3477 #5 68.91 https://nvidia.github.io/nvidia-docker/centos7/x86_64/repodata/repomd.xml: [Errno -1] Gpg Keys not imported, cannot verify repomd.xml for repo nvidia-docker
3478 #5 68.91 Trying other mirror.
3479 #5 104.0 https://developer.download.nvidia.com/compute/machine-learning/repos/rhel7/x86_64/repodata/ba5620da8b4b92fa599358bf13480976051383a7-primary.sqlite.bz2: [Errno 12] Timeout on sin: (28, 'Operation too slow. Less than 1000 bytes/sec transferred the last 30 seconds')
3480 #5 104.0 Trying other mirror.
3481 #5 134.9 https://developer.download.nvidia.com/compute/machine-learning/repos/rhel7/x86_64/repodata/ba5620da8b4b92fa599358bf13480976051383a7-primary.sqlite.bz2: [Errno 12] Timeout on sin: (28, 'Operation too slow. Less than 1000 bytes/sec transferred the last 30 seconds')
3482 #5 134.9 Trying other mirror.
3483 #5 166.0 https://developer.download.nvidia.com/compute/machine-learning/repos/rhel7/x86_64/repodata/ba5620da8b4b92fa599358bf13480976051383a7-primary.sqlite.bz2: [Errno 12] Timeout on sin: (28, 'Operation too slow. Less than 1000 bytes/sec transferred the last 30 seconds')
3484 #5 166.0 Trying other mirror.
3485 #5 197.0 https://developer.download.nvidia.com/compute/machine-learning/repos/rhel7/x86_64/repodata/ba5620da8b4b92fa599358bf13480976051383a7-primary.sqlite.bz2: [Errno 12] Timeout on sin: (28, 'Operation too slow. Less than 1000 bytes/sec transferred the last 30 seconds')
3486 #5 197.0 Trying other mirror.
3487 #5 228.2 https://developer.download.nvidia.com/compute/machine-learning/repos/rhel7/x86_64/repodata/ba5620da8b4b92fa599358bf13480976051383a7-primary.sqlite.bz2: [Errno 12] Timeout on sin: (28, 'Operation too slow. Less than 1000 bytes/sec transferred the last 30 seconds')
3488 #5 228.2 Trying other mirror.
I'm experiencing this kind of connectivity issues from nvidia repos (e.g. cuda and machine-learning) probably monthly. This server is located in China and my understanding is that the repo is backed by a glitchy CDN.
Same repo operations has way higher rate of success in US servers.