[Bug 35768] Missing file logs at far too high of log level

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=35768

Todd Vierling  changed:

   What|Removed |Added

 Status|RESOLVED|VERIFIED

--- Comment #13 from Todd Vierling  ---
(In reply to ysdev from comment #12)
> > In all seriousness: I can't do anything to fix or help a 404.
> 
> You're approaching this thing from a very narrow minded point of view. For
> you, 404 errors are apparently forced on your server from someone on the
> outside, possibly even with ill intent.
> 
> From my perspective, 404 errors happen when I link to files that don't
> exist, and so the huge majority of 404 errors I'm seeing are caused by *me*,
> until I fix my site.

And you have a way to get them anywhere you want, without them being logged to
error_log at the same level as major server problems: mod_log_config. You can
even use CustomLog's expression support to log only 404s to a file of their
own, if you prefer. And you get to choose how they're logged, including
translated path and everything else you could need.

Or just set the error log level to info if you want all content related errors
(there are more than just not-found errors logged at this level, and by the
same argument above, you may want to see those too).

The problem fixed by this bug is that major server errors and content-missing
errors had the same log level -- so the former could be drowned out by the
latter, or the former would disappear completely if the error log level were
tuned to avoid warning level messages.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 35768] Missing file logs at far too high of log level

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=35768

--- Comment #12 from ys...@tm-company.de ---
Re Jay Freeman/#9:

> If a user wanted to mess with me, they'd just start requesting 404s off my 
> server until I run out of disk space.

How exactly does the requested patch deny them the ability to do so? Best case
scenario, it slows them down by roughly 30%, unless you omit all 404 requests
from the access log as well.

> In all seriousness: I can't do anything to fix or help a 404.

You're approaching this thing from a very narrow minded point of view. For you,
404 errors are apparently forced on your server from someone on the outside,
possibly even with ill intent.

>From my perspective, 404 errors happen when I link to files that don't exist,
and so the huge majority of 404 errors I'm seeing are caused by *me*, until I
fix my site.

I'd agree that the ratio is likely to have changed over time because all kinds
of clients request all kinds of files nowadays that may or may not exist, in
order to provide browser, platform, and crawler dependent content.

> do you actually pay attention to them, and if you do, what do you do when you 
> see them? What is your tactical response to seeing a 404? Does anyone out 
> there actually want this in their logs? :(

Yes, I absolutely do.

Like I said, most of these errors are caused because *I* screwed up (or "my"
web developers/designers etcetera), and thus *I* need to fix them by correcting
broken links/image tags/script sources/background image urls.

404 errors that are caused by outside parties (apple-touch-icon foo and so on),
well, let me just ignore them.

IMHO, a better solution would have been to introduce new log levels, while
making sure that the default behavior remains unchanged; maybe by splitting
errors into error and http_error or some such.

I guess now that the change has been implemented (5 years ago, I know... but I
just came across this for the first time), there's little hope of getting the
previous behavior restored.

Anyway, the lack of documentation just cost me 3 hours of my life, so I'd have
appreciated if this had been reflected in the 2.4 documentation - at first I
honestly thought my recent Debian 8 upgrade had somehow broken Apache's error
logging in a very weird way and checked all kinds of stuff before searching the
web.

Bug report regarding the documentation: [1]

[1] https://bz.apache.org/bugzilla/show_bug.cgi?id=60247

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60244] mod_http2 causes serialized requests if there are parallel long running requests on the same connection

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60244

--- Comment #4 from Stefan Eissing  ---
The problem is that you run 'prefork' as your mpm. In prefork there is only one
connection/request per process. So, when you browser reuses the h2 connection,
it ends up in the very same process that is working on the longpoll.

What can you do?
1. Disable http2
2. Switch to mpm_event
3. Enable more H2Workers per process (the default on prefork is 1, since
prefork is intended to not have multithreading).

2 would be my recommendation. 3 might not work if you really need prefork. 1
makes me sad.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60244] mod_http2 causes serialized requests if there are parallel long running requests on the same connection

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60244

--- Comment #3 from Reno Reckling  ---
Created attachment 34368
  --> https://bz.apache.org/bugzilla/attachment.cgi?id=34368=edit
loglevel trace8 of a session that has the problem

This shows the following series of event:

Request to /test/first.html in tab #1, which loads the basic html and some
javascript.
The javascript of first.html then does an ajax request to /test/longpoll.php
which hangs for 30 seconds.
This phase is clearly visible because of the 30 second long period of:
OpenSSL: I/O error, 5 bytes expected to read on BIO#558423e82670

2 seconds after the longpoll.php request is initiated, i open another tab #2 to
/test/second.html. This one does not really show up in the log just yet.

After the longpoll in Tab #1 finished, the log says:
AH02034: Subsequent (No.3) HTTPS request received for child 2 (server
exi.wthack.de:443)

Which goes to /test/second.html and finished immediately.

Anything else i could provide to help solve this?

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60244] mod_http2 causes serialized requests if there are parallel long running requests on the same connection

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60244

Eric Covener  changed:

   What|Removed |Added

 Status|NEW |NEEDINFO

--- Comment #2 from Eric Covener  ---
loglevel trace8 of the situation pls?

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60242] Concurrent Login Sessions - Apache 2.4.20

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60242

Eric Covener  changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |INVALID

--- Comment #1 from Eric Covener  ---
Bugzilla is for reporting bugs in the server, not for support or configuration
assistance. I'd suggest a mailing list or serverfault.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60244] mod_http2 causes serialized requests if there are parallel long running requests on the same connection

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60244

--- Comment #1 from Reno Reckling  ---
Additional insight:

This happens _only_ during requests that involve php.

If i have a long running download that shares a connection with another static
file requests, everything is fine.
Could it be that mod_php7 somehow gets an exclusive lock on the connection
except just on the stream?

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60244] mod_http2 causes serialized requests if there are parallel long running requests on the same connection

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60244

e-apa...@wthack.de changed:

   What|Removed |Added

 CC||e-apa...@wthack.de

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60244] New: mod_http2 causes serialized requests if there are parallel long running requests on the same connection

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60244

Bug ID: 60244
   Summary: mod_http2 causes serialized requests if there are
parallel long running requests on the same connection
   Product: Apache httpd-2
   Version: 2.4.23
  Hardware: PC
OS: Linux
Status: NEW
  Severity: normal
  Priority: P2
 Component: mod_http2
  Assignee: bugs@httpd.apache.org
  Reporter: e-apa...@wthack.de

Hi, I just tried to move my webserver to use http2 and discovered the following
phenomenon.

This happens in chrome as well as in firefox with apache2 2.4.23.

If I open one tab to my http2 enabled domain (using h2) and it does background
ajax long-polling to a php script on the server (The script just does
sleep(30)).
While the longpoll is running, if i try to open a new tab to the same server,
the new request gets stuck in the "pending" state until the longpoll is
finished.
The second request can be against a static .html file (so no php processing)
and still gets stuck.
If I disable http2 for this setup, the requests finish in parallel just fine.

It seems to me that I somehow cause all requests to be serialized for the same
connection but different streams.
This does NOT happen for different connections, so longpoll in chrome and
second request in firefox is fine.

List of my used modules:
core.c, http_core.c, mod_access_compat.c, mod_actions.c, mod_alias.c,
mod_auth_basic.c, mod_authn_core.c, mod_authn_file.c, mod_authz_core.c,
mod_authz_groupfile.c, mod_authz_host.c, mod_authz_unixgroup.c,
mod_authz_user.c, mod_autoindex.c, mod_cgi.c, mod_dav.c, mod_dav_fs.c,
mod_deflate.c, mod_dir.c, mod_env.c, mod_filter.c, mod_headers.c, mod_http2.c,
mod_include.c, mod_info.c, mod_log_config.c, mod_logio.c, mod_mime.c,
mod_negotiation.c, mod_php7.c, mod_proxy.c, mod_proxy_http.c, mod_reqtimeout.c,
mod_rewrite.c, mod_setenvif.c, mod_so.c, mod_socache_shmcb.c, mod_ssl.c,
mod_status.c, mod_unixd.c, mod_version.c, mod_watchdog.c, mod_wsgi.c, prefork.c

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 59660] mod_macro: Control on bad nesting

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=59660

Marc Stern  changed:

   What|Removed |Added

  Attachment #34326|0   |1
is obsolete||

--- Comment #4 from Marc Stern  ---
Created attachment 34365
  --> https://bz.apache.org/bugzilla/attachment.cgi?id=34365=edit
Final version compiling on Windows/Linux

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60242] Concurrent Login Sessions - Apache 2.4.20

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60242

Shaggy  changed:

   What|Removed |Added

 CC||shirgan.parshotam@transnet.
   ||net

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org

[Bug 60242] New: Concurrent Login Sessions - Apache 2.4.20

2016-10-12 Thread bugzilla 
https://bz.apache.org/bugzilla/show_bug.cgi?id=60242

Bug ID: 60242
   Summary: Concurrent Login Sessions - Apache 2.4.20
   Product: Apache httpd-2
   Version: 2.4.20
  Hardware: Other
OS: Windows Server 2003
Status: NEW
  Severity: normal
  Priority: P2
 Component: Other Modules
  Assignee: bugs@httpd.apache.org
  Reporter: shirgan.parsho...@transnet.net

Good day Apache,

Trust you are well, we are currently utilizing apache version 2.4.20 (Windows
Server 2003 OS) on our production environment as a load balancer, we have
external users connecting via an open URL link which passes through an Palo
Alto Firewall to the load balancer. The System we are using is Navis Sparcs N4
version 2.6.19 as our Terminal Operating System, the Apache Load Balancer
fetches N4 connections from 5 Nodes configured on the Load Balancer. 

The issue we are experiencing is that there are Concurrent Login Sessions (up
to 188 logins in a second) at certain times of the day which impacts the CPU
Usage of our Database (up to 97% CPU Usage), as a result the performance of our
Navis TOS degrades and the Terminal users are affected to a point that the
system cannot process transactions.  

Is there a solution or fix in place that will not allow or block the Apache
Load Balancer from fetching Concurrent Logins from the Navis Application. 

Please feel free to contact or email me for further information. 

thank you 

Shirgan

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org