When we close a conn due to bad CERTS cell, we complain about stuff in our outbuf
While looking at bug #4371 (moved), here's what my client said:
Nov 01 04:30:45.000 [warn] Certificate not yet valid: is your system clock set incorrectly?
Nov 01 04:30:45.000 [warn] (certificate lifetime runs from Nov 1 12:15:08 2011 GMT through Oct 31 12:15:08 2012 GMT. Your time is Nov 01 08:30:45 2011 GMT.)
Nov 01 04:30:45.000 [notice] We stalled too much while trying to write 512 bytes to address "120.50.40.184". If this happens a lot, either something is wrong with your network connection, or something is wrong with theirs. (fd 14, type OR, state 7, marked at command.c:995).
Looks like we had already written our netinfo cell to the outbuf, but hadn't flushed it. Since I'm a client and this was an OR connection, it gets to be severity notice. Fun.
For the client side, I think this bug might be resolved by the proposed fix to #4361 (moved).
But my directory authority experiences something similar: Nov 01 06:04:10.000 [info] conn_close_if_marked(): Conn (addr "120.50.40.184", fd 1609, type OR, state 7) marked, but wants to flush 916 bytes. (Marked at command.c:995) Nov 01 06:04:10.000 [info] conn_close_if_marked(): We stalled too much while trying to write 916 bytes to address "120.50.40.184". If this happens a lot, either something is wrong with your network connection, or something is wrong with theirs. (fd 1609, type OR, state 7, marked at command.c:995). I wonder what's sitting in its outbuf.