| Summary: | bitbake generates zombie Parser processes (hard to reproduce) | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] BitBake | Reporter: | Yi Fan Yu <yf3yu> |
| Component: | bitbake | Assignee: | Randy MacLeod <randy.macleod> |
| Status: | RESOLVED WORKSFORME | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium+ | CC: | poky.bs.watcher, poky.watcher, randy.macleod, yf3yu |
| Version: | unspecified | ||
| Target Milestone: | 4.1 M2 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |
|
Description
Yi Fan Yu
2021-04-07 14:20:36 UTC
10061 might be related Need to figure out at which bitbake cycle are we getting stuck/hanging/needing to rm the lock file. Encountered a similar issue I believe:
```
yyu1@yow-lpggp3:/yow-lpggp32/yyu1/b/build$ MACHINE=intel-x86-64 bitbake lib32-clang-native
WARNING: You have included the meta-openstack layer, but 'openstack' has not been enabled in your DISTRO_FEATURES. Some bbappend fil.
Loading cache: 100% |#################################################################################################| Time: 0:00:01
Loaded 9617 entries from dependency cache.
Keyboard Interrupt, closing down...
NOTE: No reply from server in 30s
Timeout while waiting for a reply from the bitbake server (60s)
```
```
yyu1@yow-lpggp3:/yow-lpggp32/yyu1/b/build$ ls *ock
bitbake.lock bitbake.sock
```
```
lsof bitbake.lock
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
Cooker 35965 yyu1 3u REG 8,33 6 15328396296 bitbake.lock
Parser-2 36082 yyu1 3u REG 8,33 6 15328396296 bitbake.lock
```
```
yyu1@yow-lpggp3:/yow-lpggp32/yyu1/b/build$ ps -aux | grep Z
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
yyu1 24750 0.0 0.0 13076 1052 pts/19 S+ 16:00 0:00 grep Z
yyu1 36083 0.0 0.0 0 0 ? Z 15:34 0:00 [Parser-3] <defunct>
```
Can't kill parser3, but the lock files do disappear if I kill parser-2 with
kill `kill -s SIGTERM 36082`
Reproducible on this particular build server, but not reproducible on my personal machine. Difference is that the system makes a slight pause after seeing this ``` Loading cache: 100% ###################################################################################################################################################| Time: 0:00:00 Loaded 1438 entries from dependency cache. ``` if you hit controlc right after "Loaded 1438 ...", the bitbake server will stop responding. only reproduced on poky today, not on oe-core or wrlinux.
yyu1@yow-lpggp3:/yow-lpggp32/yyu1/yocto/b_poky/build$ bitbake core-image-minimal
WARNING: You have included the meta-virtualization layer, but 'virtualization' has not .
Loading cache: 100% |####################################################| Time: 0:00:00
Loaded 4779 entries from dependency cache.
Keyboard Interrupt, closing down...
NOTE: No reply from server in 30s
Second Keyboard Interrupt, stopping...
NOTE: No reply from server in 30s
Traceback (most recent call last):
Could you put a time.sleep(5) in the code to isolate the exact spot we need the Ctrl+C to make it easier to reproduce? I spent some time adding time.sleep to locations. And I have given up trying. it sometimes stops responding and it happens at different steps... usually it is able to recover. Sorry for not being helpful Added Yi Fan who may want to work on this bug now that he's got some time available. Richard fixed a parse time race windown and shortened the lifespan of python zombies. Re-open if seen again. |