besu --help hangs, SIGSEGV in child process

[thorstenhirsch]

Running besu 24.6.0 on RHEL 8 doesn't seem to work reliably. I only want to call besu --help, but it hangs most of the time (no output at all). Only sometimes the call runs successful.

I've traced the problem with strace and whenever the call hangs, it's keeps doing these things in a loop:

[pid  8217] getrusage(RUSAGE_THREAD, {ru_utime={tv_sec=0, tv_usec=106916}, ru_stime={tv_sec=0, tv_usec=1916}, ...}) = 0
[pid  8217] futex(0x7fb440115efc, FUTEX_WAIT_PRIVATE, 0, {tv_sec=0, tv_nsec=49999413} <unfinished ...>
[pid  8219] <... futex resumed>)        = -1 ETIMEDOUT (Connection timed out)
[pid  8219] futex(0x7fb447918ca8, FUTEX_WAKE_PRIVATE, 1) = 0
[pid  8219] futex(0x7fb447918cf8, FUTEX_WAIT_PRIVATE, 0, {tv_sec=0, tv_nsec=49999813} <unfinished ...>
[pid  8217] <... futex resumed>)        = -1 ETIMEDOUT (Connection timed out)
[pid  8217] futex(0x7fb440115ea8, FUTEX_WAKE_PRIVATE, 1) = 0
[pid  8217] getrusage(RUSAGE_THREAD, {ru_utime={tv_sec=0, tv_usec=106948}, ru_stime={tv_sec=0, tv_usec=1916}, ...}) = 0
[pid  8219] <... futex resumed>)        = -1 ETIMEDOUT (Connection timed out)
[pid  8217] futex(0x7fb440115efc, FUTEX_WAIT_PRIVATE, 0, {tv_sec=0, tv_nsec=49999618} <unfinished ...>
[pid  8219] futex(0x7fb447918ca8, FUTEX_WAKE_PRIVATE, 1) = 0
[pid  8219] futex(0x7fb447918cf8, FUTEX_WAIT_PRIVATE, 0, {tv_sec=0, tv_nsec=49999928}) = -1 ETIMEDOUT (Connection timed out)
[pid  8217] <... futex resumed>)        = -1 ETIMEDOUT (Connection timed out)
[pid  8219] futex(0x7fb447918ca8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid  8217] futex(0x7fb440115ea8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>

It seems like besu wants to connect to something that is not there. And the cause might be:

[pid  3979] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x14} ---

So there's one child process that had run on SIGSEGV. Maybe this child would have provided the server, to which the looping besu process wants to connect to.

The system is an up-to-date RHEL 8.10 (x64) with kernel 4.18 running as a virtual machine with 2 vCPU and 16GB RAM. Java version is OpenJDK 21.0.3+9.

[thorstenhirsch]

The SIGSEGV does not seem to cause besu to hang, because I also see the SIGSEGV on other Linux machines, where besu runs successful. So the question is: What else is causing besu to hang on this machine?

edit: Older versions of besu also keep hanging on this machine, I reproduced it with all versions down to 24.1.2. Downgrading to OpenJDK 17 had no effect. I increased "open files" (ulimit, hard+soft) from 1024 to 4096, but the problem still remains the same.

[thorstenhirsch]

No solution, yet, but at least a bit more info:

• when besu hangs it doesn't hang forever; it hangs for ~5min before running successfully
• in strace we see that besu is in the loop for ~5min and then, when it is finally able to exit the loop, the 1st thing it does is reading this file: /sys/fs/cgroup/memory/user.slice/user-8300.slice/session-1.scope/memory.limit_in_bytes

So is it a problem accessing the sys fs that keeps besu in the loop for ~5min?

[siladu]

Hi @thorstenhirsch thanks for the report.

I just tested this on a fresh Azure box with 2 CPU and 8GB running RHEL 8, java 21 and besu 24.6.0 and don't get the hang.

uname -a
Linux besu-rhel8 4.18.0-425.3.1.el8.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux

java -version
openjdk version "21.0.3" 2024-04-16 LTS
OpenJDK Runtime Environment (Red_Hat-21.0.3.0.9-1) (build 21.0.3+9-LTS)
OpenJDK 64-Bit Server VM (Red_Hat-21.0.3.0.9-1) (build 21.0.3+9-LTS, mixed mode, sharing)

time ./bin/besu --help
...
real    0m4.832s
user    0m8.690s
sys 0m0.297s

Are you seeing a similar hang for any other besu commands? Another one you could try is besu --version

 time ./bin/besu --version
besu/v24.6.0/linux-x86_64/openjdk-java-21

real    0m4.222s
user    0m7.597s
sys 0m0.296s

Any problems running besu itself?

[siladu]

It might be interesting to try the docker install as well to compare https://besu.hyperledger.org/public-networks/get-started/install/run-docker-image

Can you provide any details on how you installed besu? Also, anything else that is running on the machine that might be relevant? Any custom Java/JVM settings?

[siladu]

Hi @thorstenhirsch, since we couldn't recreate and haven't heard back, we will close this issue in a couple of days.