Failed to get file information for file descriptor 11: Bad file descriptor about apptainer HOT 8 CLOSED

if ssh to the server, all work OK.
we were watching the /var/log/messages while testing on the rstudio web gui. whenever that "bad file descriptor on the web interface" occurs, the overlay lowerpath message is displayed on the server .

from apptainer.

here are the captured the strace output (the complete output is attached as file):
access("/lmod/apps/apptainer/1.2.2/bin/apptainer", R_OK) = 0
rt_sigprocmask(SIG_BLOCK, [INT CHLD], [], 8) = 0
lseek(255, -1, SEEK_CUR) = 105
clone(child_stack=NULL, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x152f477bde50) = 2385409
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigaction(SIGINT, {sa_handler=0x558c001f1c60, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x152f46dfeb50}, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x152f46dfeb50}, 8) = 0
wait4(-1, ERROR : Failed to get file information for file descriptor 11: Bad file descriptor
[{WIFEXITED(s) && WEXITSTATUS(s) == 1}], 0, NULL) = 2385409
rt_sigaction(SIGINT, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x152f46dfeb50}, {sa_handler=0x558c001f1c60, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x152f46dfeb50}, 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=2385409, si_uid=90023, si_status=1, si_utime=2, si_stime=6} ---
wait4(-1, 0x7ffcc1699510, WNOHANG, NULL) = -1 ECHILD (No child processes)
rt_sigreturn({mask=[]}) = 0
read(255, "\n", 106) = 1
read(255, "", 106) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
close(5) = 0
exit_group(1) = ?
+++ exited with 1 +++

from apptainer.

I uploaded the entire strace output as a file: RStudio_strace_output.txt

from apptainer.

Is there any way you can reproduce it outside of Posit Workench? It looks like installing that is an awful lot to ask someone who doesn't have it.

That strace is not helpful because it doesn't actually trace apptainer, it traces "bedtools" and does not follow to subprocesses. A problem with using the strace -f option however is that typically it hangs on apptainer. I suggest renaming your libexec/apptainer/starter command and replacing it with a shell script that runs strace on the starter, and saving that output. The script probably needs to have full paths in it to strace and to starter. I also recommend the -s1024 option on strace.

from apptainer.

there are no problems if we ssh to the server. everything just works fine. I have given this link to the Posit support and hopefully they can help you to install a rstudio. it is quite simple, just get the rpm and a temp license, then it should work.

from apptainer.

We have resolved the “bad file descriptor” issue. it was the hyperthreading caused the problem. we have enabled and disabled logic processors a few times and the error is reproducible.

from apptainer.

The issue has been solved and can be closed. I summarised the issue and the resolution below.

The issue only appears on the RStudio Server (Posit Workbench) when you use Apptainer in the R Console or the R Terminal, and only if you have Hyperthreading enabled. Intermittently, the error message ERROR : Failed to get file information for file descriptor 11: Bad file descriptor will appear. It took us some time and effort to get to the bottom of it.

Thank you for your help and for looking into this!

from apptainer.

I would say you found a workaround, although the problem was never really understood. Oh well, since you no longer have the problem, I'm closing the issue as requested.

from apptainer.