Hi, 2014-09-03 21:46 GMT+02:00 Willy Tarreau <[email protected]>:
> Hi Olivier, > > That's intentional, and I'll reject any patch to document it. It's a > debugging feature whose format changes between versions or for any > reason we see fit when facing a new problem. This can only be used > with the code at hand. > OK I understand this point of view. Maybe stating this clearly "to have details on fields, just look at the source code, ie src/dumpstats.c" :p > The real issue I am investigating here is the "age" field. Here it is 19 > > days, should I understand that the connection was open 19 days ago ? > > Yes absolutely. > > > Maybe I'm missing something, but my HAProxy process was launched a few > > minutes ago (yet handling over an old process with -sf), so I was > expecting > > new connections on this process. Maybe I misunderstood haproxy reloading > > behaviour over TCP connections. > > Please send a "show info" there, it is possible that for any reason > you're connected to the old process (eg: if you got this from an > automated tool which stays connected to this socket). I'm not seeing > any reason for a session's age to be wrong otherwise. > No, I connected manually to the socket. The old process was still alive though. Unfortunately I was forced to fix my session issue and killed all haproxy process and relaunched it cleanly. I cannot observe this behaviour anymore. Thanks for the information, Olivier
