-
Notifications
You must be signed in to change notification settings - Fork 2.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
lnd v0.12.1-beta consumes 3-4 GB against 1-2 GB of lnd v0.11 #5066
Comments
Here's some profiles and logs. |
@gkrizek have you tweaked any settings related to pinned syncers or the historical sync interval (so gossip sync related params)? |
@Roasbeef I can say for sure when I get back to my computer but I don't remember ever adjusting gossip settings. |
@Roasbeef I just confirmed I have no settings for gossip. Is there additional information I can give for investigation? |
Adding some more info here. This node has about 400 channels. I have another node (running 0.12.1 as well) that has 5 channels and doesn't experience this problem. So I wonder if the problem gets worse with number of channels. I also tried to restart this node and it cleared out the memory obviously but CPU shot right back up and you can see memory slowly starting to climb. |
Sweet! I'll try it out tonight. |
You can provide us with a heap profile the same way as the other profiles through |
@wpaulino Here's a PNG. Let me know if |
Well, everything was running fine until I restarted the node because the memory was almost out. After a restart CPU is back to being really high again. No changes to the code (still running the patch for #5068).
|
@gkrizek could you provide the raw profile files instead for both heap and CPU? |
@wpaulino probably stupid question, do I need to password protect the heap snapshot? |
@gkrizek I don't think so, but feel free to DM it over Slack just to be safe. |
The optimizations laid out in #5075 should address some of the heap usage you're seeing w.r.t keeping blocks around in memory as well. |
Following up here. CPU seems fine but every node I have connecting to btcd seems to be leaking memory. (I don't have any connected to bitcoind so can only compare to neutrino). I've attached a heap snapshot of a node that's been growing. |
@Talkless we have several fixes in the upcoming release that should address this and the high CPU usage also commonly seen. |
Closing for now since 0.13.1 has been released, we've received reports that prior memory blow ups due to p2p activity have been remedied. 0.14 will have more optimizations that reduce burst GC activity as well. |
I noticed the strange thing.
I upgraded 5 servers to LND v0.12.1-beta and all 5 servers have the lnd processes with 3-4 GB resident memory
Other servers with LND v0.11.* have 1-1.5 GB same resident memory
Is it OK for the same version?
The text was updated successfully, but these errors were encountered: