I’ve not been very active on pants development as ...
# development
c
I’ve not been very active on pants development as of late, now getting back into it a bit I notice that
pantsd
keeps dying rather frequently for me due to: 🧵
1
Copy code
09:40:40.01 [INFO] notify invalidation: cleared 0 and dirtied 0 nodes for: {"src/python/pants/engine/internals/parser_test.py"}
09:40:40.01 [INFO] notify invalidation: cleared 0 and dirtied 0 nodes for: {"src/python/pants/engine/target.py"}
09:40:40.01 [INFO] notify invalidation: cleared 0 and dirtied 0 nodes for: {"src/python/pants/engine/internals/parser.py"}
09:40:40.01 [INFO] notify invalidation: cleared 1 and dirtied 29 nodes for: {"src/python/pants/engine/internals/parser_test.py"}
09:40:40.01 [INFO] notify invalidation: cleared 1 and dirtied 31 nodes for: {"src/python/pants/engine/target.py"}
09:40:40.01 [INFO] notify invalidation: cleared 1 and dirtied 31 nodes for: {"src/python/pants/engine/internals/parser.py"}
09:40:40.07 [ERROR] saw filesystem changes covered by invalidation globs: content changed (Digest('2a5b40046a8252018ab697eb9fa0784c87ce3070eb1b973bb97b17c0176427b0', 336) fs Digest('31505d0ccbc0ccafa49495cfb41316a1b218e1f460c10782dc0b12c342ef7ab9', 336)). terminating the daemon.
09:40:40.51 [ERROR] service failure for <pants.pantsd.service.scheduler_service.SchedulerService object at 0x15a4f5430>.
09:40:40.52 [INFO] Waiting for ongoing runs to complete before exiting...
don’t recall the daemon having to quit just because there where changed files during a run..?
b
Yes it does if it the path matches invalidation globs. Specifically these are paths for files that make up the code running the daemon
🤦 1
👀 1
n
Not sure if this is the same problem, but I've been making a few changes to the docker backend files only. After a single change to a source file I get
There is no pantsd metadata at /Users/rhys.madigan/repos/pants/.pids/559bbfd2e755/pantsd.
Modifying test files seems ok.
b
Yes, if you're modifying any part of pants, including the backends it has to restart. I assume that message comes from the native client where the hash is probably the hash of the files in the invalidation glob. Just a guess
b
The "There is no pantsd metadata" messages are apparently expected/innocuous: https://pantsbuild.slack.com/archives/C0D7TNJHL/p1684515293086589?thread_ts=1684497061.702139&amp;cid=C0D7TNJHL
👍 1