5

After updating from Bitcoin Core 28.1 to 29.0, I noticed an increase in memory usage.

After chain synchronization was complete, the memory usage for both 28.1 and 29.0 was as follows, approximately 10 hours after starting them. In bitcoin.conf, only txindex=1 is set.

Core VIRT RES SHR
28.1 4.3g 686m 287m
29.0 13.8g 7.4g 7.1g

Does 29.0 cache more chainstate and block data in memory?

My environment is Ubuntu 24.04.3 LTS.

Improve this question
8
  • That's very dramatic and unexpected. Can you post the few 100 first lines of debug.log from each? Commented Sep 8 at 12:04
  • Each debug.log is as follows: gist.github.com/azuchi/317c530129302fccc5d029281a4961ac gist.github.com/azuchi/a83def29b3d2c4297d0b5c17a0e7f7ef Commented Sep 8 at 12:38
  • I don't see anything unusual. However, I note that both RES and SHR have gone up by roughly the same amount, ~6.8 GiB. If it was really more actual memory usage, I'd expect to just see RES go up. However, I don't know what additional shared memory would be accounted for here. Commented Sep 8 at 13:27
  • 1
    I've started two identical VMs with Core 28 vs Core 29 in txindex mode in order to reproduce. I'll monitor memory usage during and after IBD today and tomorrow and will report. Commented Sep 8 at 15:34
  • 2
    @ShigeyukiAzuchi so it appears to be due to github.com/bitcoin/bitcoin/pull/30039. The memory usage reported is just due to LevelDB mmap caching and is not "real" RAM usage from the bitcoind process. This won't lead to an OOM, but it's unfortunately confusing. I think Pieter will post a proper answer here when he's got a minute. Commented Sep 9 at 16:34

0

Reset to default

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.