Phil Pennock wrote: > On 2007-05-12 at 10:02 -0700, Marc Perkel wrote:
>
>> I'm seeing a few of these in the log. What does it mean?
>>
>> SSL_Write error 5
>>
>
> It helps if you get the error message correct, including case, so that
> it's easier to find it; I suddenly thought of past indicators of
> accuracy and switched to a case-insensitive grep.
>
> That message means that the call to OpenSSL's SSL_write() failed and
> that the error code subsequently extracted from OpenSSL was unusual
> enough that it wasn't specially handled in the Exim code. Exim reports
> what the error code was.
>
> In this case, assuming that the error codes haven't been significantly
> changed between whichever version of OpenSSL you have and the version I
> have (0.9.8d), error code 5 means SSL_ERROR_SYSCALL, so there was an
> underlying system call problem. Exim doesn't go into it any deeper.
>
> You'd need to provide more details, such as context, instead of just
> that to know what's going on. I suspect that something is dropping the
> TCP connection abruptly without doing a TLS shutdown; but without
> knowing the context there's no way to know that; I give it just over
> even odds that it is this cause.
>
> -Phil
>
>
Phil, thanks for your help on this. Here's what's on my logs. For what
it's worth seems to have started on May 9 and 7:30 am. Looked in the yum
logs thinking it might have been an upgrade but no upgrades at that time.
