On Mon, 7 Jan 2008, Willy Tarreau wrote:
> Hi Krzysztof,
> I'm writing to you because I found a way to implement the turn-around
> delay after a failed connection attempt and I know you're waiting for
> this :-)
> Finally I did not add any state to the server's FSM. I proceeded
> differently, with a new SN_CONN_TAR flag added to the session, I
> fake a server connection timeout when a connection error occurs.
> Initially, I wanted to reuse the contimeout as the turn-around
> timeout, but there are people using very large contimeouts because
> it was the same value which dictated the tarpit and queue timeouts
> in previous versions, so this would result in reconnection attempts
> being delayed too long. So finally I arbitrarily selected a one second
> delay to attempt a reconnection after a failure.
> If this value is considered too low by some users and too high by others,
> then we'll introduce "timeout reconnect" :-)
Are there any chances that with this patch (commit 541b5c24ca070789f2474a6c8f2398dc3877aeb5) in some, not yet clear to me situation, it is possible to get a server with a negative connection count? It happened serveral times and I connect it somehow with this patch as AFAIK there were no other changes in this area.
I have never observed such situation without this patch, but as I have not been able to debug it to much, so maybe my shot is wrong. I'll try to get some more information but maybe you have an idea?.
Krzysztof Olędzki Received on 2008/04/03 01:09
This archive was generated by hypermail 2.2.0 : 2008/04/03 01:15 CEST