Apache fills up error log with TransferLog

Question

I'm using Apache 2.4 and several virtual hosts. This is on Windows, BTW. Within each virtual host, I have the following log file configurations:

CustomLog "|bin/rotatelogs.exe -l ${SRVROOT}/logs/dev.ssl_request.%Y.%m.%d.log 86400" \
      "%t %h %{SSL_PROTOCOL}x %{SSL_CIPHER}x \"%r\" %b" env=HTTPS

ErrorLog "|bin/rotatelogs.exe -l ${SRVROOT}/logs/dev.error.%Y.%m.%d.log 86400"
TransferLog "|bin/rotatelogs.exe -l ${SRVROOT}/logs/dev.access.%Y.%m.%d.log 86400"

That would be for the DEV virtual host. The others (test/ua/etc.) are basically the same except for the ServerName and logs are named {environment}.error, etc.

Now, when I start up Apache, the error log file gets hammered with the exact same message over and over. That message is:

Incorrect number of arguments
Usage: C:\Program Files\Apache Software Foundation\Apache24\bin\rotatelogs.exe [-v] [-l] [-L linkname] [-p prog] [-f] [-t] [-e] [-n number] <logfile> {<rotation time in seconds>|<rotation size>(B|K|M|G)} [offset minutes from UTC]

Add this:

TransferLog "|C:\Program Files\Apache Software Foundation\Apache24\bin\rotatelogs.exe /some/where 86400"

or 

TransferLog "|C:\Program Files\Apache Software Foundation\Apache24\bin\rotatelogs.exe /some/where 5M"

to httpd.conf. By default, the generated name will be
<logfile>.nnnn where nnnn is the system time at which the log
nominally starts (N.B. if using a rotation time, the time will
always be a multiple of the rotation time, so you can synchronize
cron scripts with it). If <logfile> contains strftime conversion
specifications, those will be used instead. At the end of each
rotation time or when the file size is reached a new log is
started.

Options:
  -v       Verbose operation. Messages are written to stderr.
  -l       Base rotation on local time instead of UTC.
  -L path  Create hard link from current log to specified path.
  -p prog  Run specified program after opening a new log file. See below.
  -f       Force opening of log on program start.
  -t       Truncate logfile instead of rotating, tail friendly.
  -e       Echo log to stdout for further processing.

The program is invoked as "[prog] <curfile> [<prevfile>]"
where <curfile> is the filename of the newly opened logfile, and
<prevfile>, if given, is the filename of the previously used logfile.

I cannot figure out why this is happening. Also, before I put Apache on our web server, I ran the exact same setup locally on my laptop and the logging worked without error messages.

Any ideas?

Thanks.

UPDATE

As suggested by Jeff Snider, I went one step further and removed the "${SRVROOT}" and changed everything to "logs/dev.error....". I read in the Apache docs that if you don't specify a / then it defaults to SRVROOT anyway.

After I did that I started seeing some logs pop up. And, the error log was stable. But, after a few minutes it started happening again.

I even disabled ALL "TransferLog"'s in each environment and only have the CustomLog and ErrorLog. Same thing.

My current config is now:

CustomLog "|bin/rotatelogs.exe -l logs/dev.ssl_request.%Y.%m.%d.log\ 86400" \
      "%t %h %{SSL_PROTOCOL}x %{SSL_CIPHER}x \"%r\" %b" env=HTTPS

ErrorLog "|bin/rotatelogs.exe -l logs/dev.error.%Y.%m.%d.log 86400"

score 1 · Accepted Answer · answered Oct 19 '17 at 15:22

1

Try putting quotes around ${SRVROOT} in your command lines. I suspect that has spaces and those might confuse it by looking like several separate arguments.

answered Oct 19 '17 at 15:22

Jeff Snider

3,272
18
17

That actually helped but the problem still remains. See my edit above. – cbmeeks Oct 19 '17 at 15:50
Your current config on the CustomLog line reads ...`%d.log\ 86400"`... That backslash may be causing trouble now. – Jeff Snider Oct 19 '17 at 16:53
Ah, I didn't notice that. I will try that out. Thanks. – cbmeeks Oct 19 '17 at 17:19
The `log\ 86400` error seemed to do the trick. Thanks. – cbmeeks Oct 20 '17 at 14:15

Apache fills up error log with TransferLog

1 Answers1