163 lines
4.4 KiB
Markdown
163 lines
4.4 KiB
Markdown
# WSL Shared Conda Runtime
|
|
|
|
此目录用于维护当前统一的 WSL 共享 conda 运行时文档与锁定文件。
|
|
|
|
## 当前策略
|
|
|
|
当前项目不再建议为 `isce2`、`timeseries`、`pyint` 分别维护多套相互漂移的 conda 环境，而是收口为：
|
|
|
|
- `base`
|
|
- `insar_wsl_v1`
|
|
|
|
目标是让以下链路共享同一套 Python 运行时：
|
|
|
|
- ISCE2 D-InSAR
|
|
- 时序 InSAR 当前 SBAS 流程
|
|
- Gamma / PyINT 的 Python 胶水层
|
|
|
|
说明：
|
|
|
|
- Gamma 二进制本体仍然保持固定安装位置，不建议放入 conda。
|
|
- Gamma 相关的 `PATH`、`GAMMA_HOME`、脚本目录注入，统一由 `deploy/wsl/profiles/gamma_env.sh` 负责。
|
|
|
|
## 当前约定产物
|
|
|
|
本目录建议维护以下文件：
|
|
|
|
- `insar_wsl_v1.environment.yml`
|
|
人工可读、可维护的环境定义。
|
|
|
|
- `insar_wsl_v1.explicit.lock`
|
|
由 `conda list --explicit` 导出的精确锁文件。
|
|
|
|
- `insar_wsl_v1.fingerprint.json`
|
|
用于记录 Python 版本、关键包版本、生成时间等信息。
|
|
|
|
## 推荐工作流
|
|
|
|
### 1. 在 WSL 中核出现有环境
|
|
|
|
先查看现有环境：
|
|
|
|
```bash
|
|
conda env list
|
|
```
|
|
|
|
逐个导出候选环境包列表，用于做并集：
|
|
|
|
```bash
|
|
conda list -n isce2
|
|
conda list -n isce2_mintpy_v1
|
|
conda list -n pyint
|
|
```
|
|
|
|
### 2. 整理并集并创建新环境
|
|
|
|
推荐先人工整理 `insar_wsl_v1.environment.yml`，再创建环境：
|
|
|
|
```bash
|
|
conda env create -n insar_wsl_v1 -f deploy/wsl/conda/insar_wsl_v1.environment.yml
|
|
```
|
|
|
|
如果环境已存在：
|
|
|
|
```bash
|
|
conda env update -n insar_wsl_v1 -f deploy/wsl/conda/insar_wsl_v1.environment.yml --prune
|
|
```
|
|
|
|
### 3. 生成锁文件与指纹
|
|
|
|
```bash
|
|
conda env export --from-history -n insar_wsl_v1 > deploy/wsl/conda/insar_wsl_v1.environment.yml
|
|
conda list --explicit -n insar_wsl_v1 > deploy/wsl/conda/insar_wsl_v1.explicit.lock
|
|
```
|
|
|
|
```bash
|
|
python - <<'PY'
|
|
import json
|
|
import subprocess
|
|
|
|
def sh(*argv):
|
|
return subprocess.check_output(argv, text=True).strip()
|
|
|
|
payload = {
|
|
"python_version": sh("python", "--version"),
|
|
"conda_env": "insar_wsl_v1",
|
|
"isce_version": sh("python", "-c", "import isce, sys; print(getattr(isce, '__file__', 'unknown'))"),
|
|
}
|
|
print(json.dumps(payload, indent=2, ensure_ascii=False))
|
|
PY
|
|
```
|
|
|
|
将输出整理后保存为 `insar_wsl_v1.fingerprint.json`。
|
|
|
|
## .env 对齐要求
|
|
|
|
共享环境建好后，应确保 `.env` 中以下键值一致：
|
|
|
|
```env
|
|
WSL_DISTRO=Ubuntu-24.04
|
|
WSL_SHARED_CONDA_ENV=insar_wsl_v1
|
|
WSL_SHARED_PYTHON=/home/administrator/miniconda3/envs/insar_wsl_v1/bin/python
|
|
|
|
ISCE2_PYTHON=/home/administrator/miniconda3/envs/insar_wsl_v1/bin/python
|
|
TIMESERIES_ENV_NAME=insar_wsl_v1
|
|
TIMESERIES_PYTHON=/home/administrator/miniconda3/envs/insar_wsl_v1/bin/python
|
|
PYINT_WSL_PYTHON=/home/administrator/miniconda3/envs/insar_wsl_v1/bin/python
|
|
```
|
|
|
|
## 运维验证
|
|
|
|
系统启动后，运维自检中的 `wsl_runtime` 应满足：
|
|
|
|
- `shared_distro = Ubuntu-24.04`
|
|
- `shared_conda_env_name = insar_wsl_v1`
|
|
- `required_runtime_count == healthy_runtime_count`
|
|
|
|
## ISCE2 Rubbersheeting Dependency
|
|
|
|
The managed `ISCE2` `lt1_stripmap` profile enables native range and azimuth
|
|
rubbersheeting by default. ISCE2's `runRubbersheetRange.py` imports
|
|
`astropy.convolution` during that step, so the shared WSL runtime must include
|
|
`astropy`.
|
|
|
|
Check the active runtime:
|
|
|
|
```bash
|
|
/home/administrator/miniconda3/envs/insar_wsl_v1/bin/python -c "from astropy.convolution import convolve; print('astropy_ok')"
|
|
```
|
|
|
|
Install or repair the package in an existing deployment:
|
|
|
|
```bash
|
|
conda install -n insar_wsl_v1 -c conda-forge astropy
|
|
```
|
|
|
|
After changing the runtime, regenerate `insar_wsl_v1.explicit.lock` and
|
|
`insar_wsl_v1.fingerprint.json` so migrations can reproduce the same capability.
|
|
|
|
## ISCE2 Ionosphere Dependency
|
|
|
|
The managed `ISCE2` `lt1_stripmap` profile now runs the native stripmap
|
|
ionosphere path (`split spectrum`, low/high-band unwrap, and `ionosphere`)
|
|
before geocoding. That step requires both `cv2` and `scipy` in the shared WSL
|
|
runtime.
|
|
|
|
Check the active runtime:
|
|
|
|
```bash
|
|
/home/administrator/miniconda3/envs/insar_wsl_v1/bin/python -c "import cv2, scipy; print('ionosphere_ok')"
|
|
```
|
|
|
|
Install or repair the packages in an existing deployment:
|
|
|
|
```bash
|
|
conda install -n insar_wsl_v1 -c conda-forge opencv scipy
|
|
```
|
|
|
|
If you change the runtime package set, regenerate
|
|
`insar_wsl_v1.explicit.lock` and `insar_wsl_v1.fingerprint.json` so deployment
|
|
and migration records stay aligned with the actual environment.
|
|
|
|
如果健康面板显示 Python 路径不一致，说明仍然有旧环境残留在配置层或运行时注册层。
|