Apache WebServer IO wait
Hi Folks!
I linked a screenshot below where you can notice high IO time (3.5sec) at the first (web request) node, I can see this issue on almost every purepath despite of the fact there were idle webserver threads at that time. Can dynaTrace really reveal us what might have been the issue and why spent the request here 3.5sec?
I am just guessing that maybe some apache loaded module causes this issue. Do you have any other idea or past experience what should I suspect?
Thanks.
At first thanks so much for your answer!
The response size was 420kbyte without any compression. As you can see below the elapsed time was 0 sec in my case .
Any chance you have modules that process the response before it gets sent back to the caller? Minifications, Compression, Encryption, ... come to mind. The PurePath clearly show that the request itself right away gets passed to Java - so - my conclusion here is that most of the time is spent in modules that get executed after Java returns the HTTP Response - hence thinking of Minification, Compression ...
Thank you for your answer!
We used these modules, perhaps the mod_ssl caused this issue.
mod_plsql.c, mod_dms.c, ohs specific monitoring mod_weblogic.c, WL proxy mod_ssl.c, mod_context.c, ohs specific. mod_setenvif.c mod_authn_anon.c, mod_authn_file.c mod_authz_user.c, mod_auth_basic.c mod_alias.c mod_info.c mod_status.c mod_negotiation.c, mod_mime.c, mod_mime_magic.c mod_log_config.c ora_audit.c, ohs_core.c,, odl_log.c mod_so.c, http_core.c, worker.c, core.c
Hi Balazs,
Unfortunately finding out what Apache is doing internally is always a bit tricky, as we can't see into that native part. What modules are you loading? If you're using mod_status, you could look at the Apache Monitor Plugin to get more insight how your worker threads are spending their time
-Wolfgang
Hi,
If you look at the "elapsed time" column, this should show you a timeline of the nodes in the PurePath, and indicate if time was spent on the Web Server before or after the call to service() on your Application Server. This may be a clue to what caused the issue (I'd guess loadmodule instances if the elapsed time showed around 3600 ms for service() or compression/response size if the elapsed time showed around 0-100 ms, but those are not the only possibilities I think). Is there only one Synchronous Path call coming from the Web Server (i.e. if you collapse "service()" is there anything else in the PurePath)?
Feel free to share for a better analysis.
Hope that helps,
Rick B
