[raspberrypi-4-4gb-ram] [monerod] Constantly fail downloading the chain at 96-96%

[jscoobyced]

I am using Ubuntu Core 20.10 on a Raspberry Pi 4 with 4GB of RAM. I have mounted an SSD (via USB 3) on /opt and the monero I have built from sources following the README, using these settings:

cmake -DNO_AES=ON
make

Then I am running the monerod using this command as a systemd service:

/opt/monero/current/bin/monerod --data-dir /opt/monero/.bitmonerod --non-interactive --enforce-dns-checkpointing --db-sync-mode=safe:sync --prep-blocks-threads 2

I have also tried simpler commands without the --enforce-dns-checkpointing --db-sync-mode=safe:sync --prep-blocks-threads 2 part. I have also enabled in /etc/dphys-swapfile:

CONF_MAXSWAP=2048

I have tried with more swap as well.

The download process goes well and after about 30 hours, I'm getting around 96% (sometimes 97%) and then it starts failing with this error (I have removed 2020-12-11 05:39:06.496 [P2P7] INFO stacktrace beginning of each line for readability)

src/common/stack_trace.cpp:133  Exception: std::runtime_error
src/common/stack_trace.cpp:134  Unwound call stack:
src/common/stack_trace.cpp:172      [1]  0x12c) [0xaaaad8602350]:__cxa_throw+0x12c) [0xaaaad8602350]
src/common/stack_trace.cpp:172      [2]  0x6c) [0xaaaad8c0b1fc]:_Z21allocLargePagesMemorym+0x6c) [0xaaaad8c0b1fc]
src/common/stack_trace.cpp:172      [3]  0x154) [0xaaaad8c01e24]:_alloc_cache+0x154) [0xaaaad8c01e24]
src/common/stack_trace.cpp:172      [4]  0x21c) [0xaaaad8a13d3c]:_slow_hash+0x21c) [0xaaaad8a13d3c]
src/common/stack_trace.cpp:172      [5]  0x124) [0xaaaad8a01a28]:_ZN10cryptonote18get_block_longhashEPKNS_10BlockchainERKNS_5blockERN6crypto4hashEmPKS7_i+0x124) [0xaaaad8a01a28]
src/common/stack_trace.cpp:172      [6]  0x24) [0xaaaad8a01b84]:_ZN10cryptonote18get_block_longhashEPKNS_10BlockchainERKNS_5blockEmi+0x24) [0xaaaad8a01b84]
src/common/stack_trace.cpp:172      [7]  0xe4) [0xaaaad89b3168]:_ZNK10cryptonote10Blockchain21block_longhash_workerEmRKN4epee4spanIKNS_5blockEEERSt13unordered_mapIN6crypto4hashESA_St4hashISA_ESt8equal_toISA_ESaISt4pairIKSA_SA_EEE+0xe4) [0xaaaad89b3168]

Seeing the first error being around large pages memory, I tried to see what I can do about it, but couldn't. I've seen something around some huge page configuration in sysctl but I don't find any in my Ubuntu config. I have followed the https://github.com/peanutsformonkeys/raspberry-monerod/blob/master/README.md and now trying to sync again after adding the swapiness to 0. I doubt it would do but hey, I'll try all recommended settings.

Anything else I can do? What can I do to not have this large page limitation?

[moneromooo-monero]

That's odd, this ought not to cause a failure. Try running with --log-level 4 (warning: very spammy), it's likely that there will be more logs after this, and the cause of the failure is unrelated to this large page message.

[jscoobyced]

Thank you, going to try.

[moneromooo-monero]

Also, check dmesg, it might just be the OOM killer.

[jscoobyced]

The process doesn't die by the way, it keeps logging the error. It even seem to continue downloading but at much lower speed. I have restarted the download, I'll enable lvl 4 logging when it reaches 95% and see.

[moneromooo-monero]

Ah, good. The slowness is because it start PoW checks, which it could skip for most of the chain. This can be closed I think, if it's working on your end ?

[jscoobyced]

It failed again, but found the reason. I found in syslog that my SSD drive is disconnecting and reconnecting occasionally. It is attached to the RPi by a SSD HAT but maybe the monerod process is too intensive for it. I'll close this issue as not related to monerod. Thanks again for the suggestions.