Bump file descriptor rlimit to hard rlimit by default #315
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cvonelm
force-pushed
the
issue-313-bump-fd-rlimit
branch
from
b9b725a to
b30ece8
Compare
bmario
reviewed
Feb 5, 2024
src/util.cpp
Outdated
| @@ -366,4 +366,24 @@ std::string get_nec_thread_comm(Thread thread) | |||
| // If no '--' is found, fall back to the complete commandline as a name | |||
| return std::accumulate(args.begin(), args.end(), std::string("")); | |||
| } | |||
|
|
|||
| struct rlimit save_rlimit_fd() | |||
There was a problem hiding this comment.
Don't call it save, if it only saves one time and then is used as getter. Also, which or whom rlimit is saved? Maybe inherited_rlimit_fd, initial_rlimit_fd, or parent_process_rlimit_fd instead?
src/monitor/process_monitor_main.cpp
Outdated
| @@ -55,6 +55,9 @@ namespace monitor | |||
|
|
|||
| [[noreturn]] static void run_command(const std::vector<std::string>& command_and_args) | |||
| { | |||
| struct rlimit saved_rlimit = save_rlimit_fd(); | |||
| setrlimit(RLIMIT_OFILE, &saved_rlimit); | |||
There was a problem hiding this comment.
Why is it RLIMIT_OFILE instead of RLIMIT_NOFILE?
bmario
approved these changes
Feb 5, 2024
The default soft limit for the number of open file descriptors per-process in most Linux systems is 1024. This results in crashes on most HPC systems I've used recently as even simple lo2s invocations will exceed this limit with all the per-core perf_event_open calls. This microscopic soft limit is in place because select() only allows fd's < 1024. If you do not plan to use select() in your code, it is safe to bump the file descriptor limit from the soft limit to the hard limit.
cvonelm
force-pushed
the
issue-313-bump-fd-rlimit
branch
from
c572212 to
db8dfe0
Compare
|
All those calls need error handling. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The default soft limit for the number of open file descriptors per-process in most Linux systems is 1024. This results in crashes on most HPC systems I've used recently as even simple lo2s invocations will exceed this limit with all the per-core perf_event_open calls.
This microscopic soft limit is in place because select() only allows fd's < 1024. If you do not plan to use select() in your code, it is safe to bump the file descriptor limit from the soft limit to the hard limit.
We need to save and restore the old limit before we start the program under measurement however, as the resource limits are inherited by forked processes and we can not guarantee that the program under measurement does not do stupid stuff with select()