Yeastplume

  This image shows file descriptors being left open in `/proc/[pid/fd`, which when counted just happen to be at the user FD limit. To reproduce, I'd try running...

I have this reproduced after running a node fresh for a couple of weeks, we can see the file handles being left oper:  The last message in the logs...

Threads seem to be stuck trying to obtain a read lock of the header pmmr: Chain::process_block_header, thread `peer_read`, stuck waiting for exclusive lock on either header_pmmr or txhashset  `peer_write`...

Also one instance of an exclusive lock waiting for all readers to complete, grin_chain::chain::Chain::process_block_header 

In normal operations, when a node falls 5 blocks behind, header sync kicks off followed by a message that syncronization was successful. 20220824 12:49:20.586 INFO grin_servers::grin::sync::syncer - sync: total_difficulty 1965578938893570,...

Reported long ago but appears a valid concern. Has anyone verified behaviour on the current code?

Is this likely to be addressed? We're fairly tied into to hyper at this stage and I still don't believe there are any hugely compelling alternatives.

@pkariz thanks very much for clearly reporting this, I appreciate it. I've done some work since that may fix these issues, would you be able to test against the latest...

It's worth investigating this and seeing how easy it is to upgrade (as well as the hyper crate in #3717), but I wouldn't expect either of these to be trivial...

This is a node issue, not a wallet issue, but please check https://github.com/mimblewimble/grin/issues/3725. You might want to try rebuilding from the latest `pibd_impl` branch as a potential fix is in...