OK thanks, the device was a mobile, so maybe they were aborts. Would not
be sending ValueButNoName.
I also had three extra gets/db entries, which cannot happen as I use a
token, around the same time the exception happened. Strange.
xx.xx.xx.xxx - - [27/May/2023:22:16:35 +0100] "GET /
xx.xx.xx.xxx - - [27/May/2023:22:16:35 +0100] "GET /
xx.xx.xx.xxx - - [27/May/2023:22:16:35 +0100] "GET /
xx.xx.xx.xxx - - [27/May/2023:22:16:40 +0100] "GET /
Cheers
On 29/05/2023 16:52, Mark Thomas wrote:
On 29/05/2023 08:56, Greg Huber wrote:
Hello,
Looking through my logs I am seeing a few:
org.apache.catalina.connector.ClientAbortException:
java.io.IOException: Broken pipe
<snip/>
Checking previous logs for 9.0.74, there were none, and I have not
made any changes to my app. Were there any changes that would cause
these?
Nothing obvious.
It is AJP so that suggests the proxy is closing the connection before
Tomcat expects it to. That might happen for a 400 response but to be
honest, I'd expect the proxy to read the full response before closing
the connection. There was some tightening up of HTTP request header
validation that might cause a few more 400 responses but I'd be
surprised since the proxy should have rejected those requests before
forwarding them anyway.
The extra checks are for HTTP headers without names. i.e.:
: ValueButNoName
Maybe check if the client is sending such a header?
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
