httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "jb2002@pc9.org" <jb2...@pc9.org>
Subject [users@httpd] Apache/2.0.40, mod_ssl unexplainable errors logged
Date Wed, 28 Aug 2002 14:25:28 GMT
Here is my setup (for www.pc-tools.net): Apache/2.0.40 (Unix) 
mod_ssl/2.0.40 OpenSSL/0.9.6g. Running on Linux 2.4 kernel.

I am getting tons of these errors in apache's log file, sometimes 
separated by tens of minutes, other times separated by only a few 
seconds. I don't see any regularity (however I DID detect a correlation 
with non-SSL hits served, see below)

[error] Spurious SSL handshake interrupt [Hint: Usually just one of those 
OpenSSL confusions!?]

At first I thought this might be the result of people connecting to my 
SSL server. Then I found out that this is NOT the case. I firewalled off 
the https port so that nobody could reach my SSL server (the port 80 
server still gets plenty of traffic, however). For my remaining tests 
there was no SSL site access at all.

Doing tail -f I can watch the errors continue to appear. One odd thing I 
noticed is that whenever the error appears, netstat shows this local 
connection on the server, with varying port (1924, 1936, 1949)

tcp        0      0 localhost:1924          localhost:https

There are no "special" programs running that could cause this local, other 
than httpd itself. Next time the error occured, I got netstat to dump 
PID/Program name which turned out to be "-"

127.0.0.1:2259          127.0.0.1:443           TIME_WAIT   -

There is no PID or program name reported. But whenever one of those 
"Spurious SSL handshake interrupt" messages appears, this localhost to 
localhost connection has taken place (cause, or effect?)

So this leads me to believe that this error appears when a local 
connection originates from Apache back to itself. I hypothesize that when 
an httpd thread serves a number of requests and terminates/resets, this 
causes the error (when mod_ssl gets confused upon the reset). To test the 
hypothesis I dug into my logs.

Over the period of 2 days (which isn't a lot of data points) I extracted 
the time stamp from each 'Spurious' error and the time stamp from each HTTP 
request served. I tabulated both based on hour of the day, and plotted the 
results. The results seem to suggest that the trend of the errors is 
related to the trend of general HTTP traffic, which might support the idea 
of the error being caused by threads dying/reseting and pissing off 
mod_ssl.

So... what can I do to stop those "Spurious SSL handshake interrupt" errors 
from appearing? As I've shown, it IS NOT related to external SSL site 
traffic. Is this an Apache or mod_ssl bug? What is that self-initiated 
local https connection?

Any help appreciated. Regards,

Jem Berkes



---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
   "   from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org


Mime
View raw message