-
Notifications
You must be signed in to change notification settings - Fork 20.8k
Geth just keeps rewinding after crash in hash mode #31586
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
Comments
The trace is very strange because the crash but clearly, the case of a nil input should be caught above in that library function before it can panic. seems like memory corruption could manifest as a mismatch between the stored vs actual size of a slice? |
I've no idea. Machine details: OS Name Microsoft Windows 10 Pro Previous memory tests have all come back clear. I could run some more tonight and see. Is there anything I can do to recover from here? |
Whats the command line flags you used for Geth, and what's the Geth version? |
Geth 1.15.7 [Eth.Miner] [Eth.TxPool] [Eth.GPO] [Node] [Node.P2P] [Node.HTTPTimeouts] [Metrics] |
I restarted and it seems to be downloading from scratch:
At least it hasn't had any errors for a while, but it does have these The pending number goes up and down as does the eta. Synced has said 100% for 3hrs now. |
It ran for a few hours, then started erroring and eventually crashed. Odd to see out of memory when the system has over 200Gb ram free.
|
This part is related to #31508 |
@cosmicnet did you manage to run that memory test? So far best guess is the RAM is fried as @jwasinger mentioned |
Memory test came back clear. I'm about to go away for a few days, I could run a longer test when I get back, but I don't think it's the memory. I remember another go based program throwing some memory errors at one point, I ran a lot of memory tests and nothing showed up. Then in their next version the problem was fixed. I used this machine for a few things, and none of the other programs are having issues. When I've had a machine with bad ram in the past all sorts of stuff gets affected. It's looking like I'll have to start from scratch. I'm seeing odd stuff I've never seen before:
I'm hoping there is a short cut to restarting a full node rather than syncing from block 0? |
It's on hash mode, reason: @cosmicnet do you ever update to v1.15.6? There is a bug affects the hash mode, causing the state to be corrupted |
I probably was on 1.15.6 for a while. |
Saying that I can't see it in my download history. Looks like I went 1.15.1 -> 1.15.7 |
You said geth wasn't gracefully shutting down. You probably had to kill it. It should in that case print a stack trace. That would be very useful for debugging. Do you still have it? |
So the bad news is, since this node may have been running v1.15.6 at some point, the state database could be broken. I would advise to just upgrade to v1.15.7 and resync. It will take a day or so, since your node is not an archive node. |
I've only got the stack traces from the two crashes I captured. I ran another memory test and it came back clear again. |
Hi,
Geth crashed. I don't understand why, I have a lot of output.
When I started it back up it just keeps saying Block state missing, rewinding further and at this rate is heading back to 0 :(
I've stopped geth so it doesn't go any further.
Idk what to do, I don't want to start from scratch. I'm running a full node.
For me Geth had been acting oddly since the last update, Ctrl-C would start a shutdown but after giving it some time it wouldn't actually shut down.
Below is some output, but I've got more from the crash if needed.
The text was updated successfully, but these errors were encountered: